The present inventions relate to methods and systems for non-invasive measurements from the human body, and in particular, methods and systems related to detecting physiological activity from the human brain, animal brain, and/or peripheral nerves.
Measuring neural activity in the brain is useful for medical diagnostics, neuromodulation therapies, neuroengineering, and brain-computer interfacing. Conventional methods for measuring neural activity in the brain include X-Ray Computed Tomography (CT) scans, positron emission tomography (PET), functional magnetic resonance imaging (fMRI), or other methods that are large, expensive, require dedicated rooms in hospitals and clinics, and are not wearable or convenient to use.
In contrast to these techniques, one promising technique for measuring neural activity in the brain is magnetoencephalography (MEG), which is capable of non-invasively detecting neural activity in the brain without potentially harmful ionizing radiation, and without use of heavy or large equipment. Thus, MEG-based neural activity measurement systems can be scaled to wearable or portable form factors, which is especially important in brain-computer interface (BCI) applications that require subjects to interact freely within their environment. MEG operates under the principle that time-varying electrical current within activated neurons inherently generate magnetic signals in the form of a magnetic field that can be detected by very sensitive magnetometers located around the head.
Measuring the small magnetic fields emanating from the brain, and doing so non-invasively (without surgically penetrating the skin and bone of the head) and doing so with high spatial and temporal resolution, is difficult. The magnetic fields produced by the brain are small, and they are smaller still by the time they propagate out past the skull and the skin surface of the head. In comparison, the magnetic field emitted from various outside magnetic sources in the environment, including from global sources, such as the Earth's magnetic field, and from localized sources, such as electrical outlets and sockets, electrical wires or connections in the wall, and everyday electrical equipment in a home, office, or laboratory setting, far exceed the strength of the magnetic signals generated in the brain by many orders of magnitude, and has a distribution in space and time that is not known a-priori. Hence, it is a difficult challenge to extract the small desired signal from the brain, and to discriminate it from much larger unwanted magnetic field signals from the rest of the user's natural environment.
One type of system that can be used for MEG is a Superconductive Quantum Interference Device (SQUID), which is sensitive enough to measure magnetic fields as small as 5×10−18 Tesla, which can be compared to magnetic fields resulting from physiological processes in animals, which may be in the range of 10−9 to 10−6 Tesla. However, SQUIDs rely on superconducting loops, and thus require cryogenic cooling, which may make it prohibitively costly and too large to be incorporated into a wearable or portable form factor. Thus, neural activity measurement systems that utilize SQUIDs may not be appropriate for BCI applications.
Optically pumped magnetometers (OPMs) have emerged as a viable and wearable alternative to cryogenic, superconducting, SQUID-based MEG systems, and have an advantage of obviating the need for cryogenic cooling, and as a result, may be flexibly placed on any part of the body, including around the head, which is especially important for BCI applications. Because cryogenic cooling is not required, OPMs may be placed within millimeters of the scalp, thereby enabling measurement of a larger signal from the brain (brain signals dissipate with distance), especially for sources of magnetic signals at shallow depths beneath the skull, as well as providing consistency across different head shapes and sizes.
OPMs optically pump a sample (usually a vapor formed of one of the alkali metals (e.g., rubidium, cesium, or potassium) due to their simple atomic structure, low melting point, and ease of pumping with readily available lasers) with circularly polarized light at a precisely defined frequency, thereby transferring polarized light to the vapor, and producing a large macroscopic polarization in the vapor in the direction of the light (i.e., the alkali metal atoms in the vapor will all have spins that are oriented in the direction of the light) that induces a magnetically sensitive state in the vapor. Once this magnetically sensitive state is established, polarized light is no longer transferred to the vapor, but instead, passes transparently through the vapor. In the presence of an ambient magnetic field, the spin orientation (or precession) of the alkali metal atoms in the optically pumped vapor will uniformly change, thereby disrupting the magnetically sensitive state, which is then subsequently reestablished by the transfer of the polarized light to the vapor. Because the transmission of light through the vapor varies as the spin precession of the alkali metal atoms in the vapor (and thus the magnetically sensitive state) changes in response to changes in the ambient magnetic field, the transmission of light (either the pumping light or a separate probe light) through the vapor represents a magnetic field-dependent signal (i.e., a MEG signal) that may be detected, thereby providing a measure of magnitude changes in the magnetic field.
To maintain the magnetically sensitive state of the vapor, it is important that spin relaxation due to spin exchange collisions be suppressed. In low magnetic fields (<10 nT), spin relaxation due to spin exchange collisions can be suppressed greatly, and thus, some OPMs are operated as zero-field magnetometers or Spin Exchange Relaxation Free (SERF) OPMs (referred to as “SERF OPMs”), thereby allowing for very high magnetometer sensitivities. Furthermore, because OPM measurements can be quite sensitive to low-frequency noise, the polarization of the vapor may be modulated to move the MEG signal away from the low-frequency end of the spectrum. SERF OPMs typically amplitude modulate the vapor polarization using magnetic coils that generate oscillating magnetic fields that vary at a frequency (e.g., 2000 Hz) much greater than the relaxation rate of the vapor (approximately 100 Hz). The amplitude modulated MEG signal can then be demodulated using lock-in detection to recover the MEG signal.
Although SERF OPMs allow for very high magnetometer sensitivities, they have a small dynamic range and bandwidth compared to SQUIDs, and can thus only operate in small magnetic fields (tens of nT, and often lower, to stay in the linear range of the OPMs). This becomes problematic when attempting to detect a very weak neural activity-induced magnetic field from the brain against an outside magnetic field.
For example, referring to
The approximate operating range of a SERF OPM (i.e., the range in which the metallic alkali vapor resonates) extends from below 1 fT up to 200 nT. Outside of this range, the metallic alkali vapor in the OPM loses sensitivity to magnetic fields. In contrast, the approximate operating range of a less sensitive sensor, such as a flux gate magnetometer, extends from around 100 fT to close to 100 μT. Thus, in contrast to flux gate magnetometers, the limited dynamic range of a SERF OPM presents a challenge in measuring signals having a high dynamic range, e.g., approximately 2×1010, which corresponds to the ratio of the lower range magnitude of the MEG signal (approximately 5 fT) to the higher range magnitude of the outside magnetic field (approximately 100 μT).
Thus, to take advantage of SERF OPMs for MEG, the outside magnetic field must be suppressed to near-zero. Otherwise, the SERF OPM cannot operate. One conventional technique for suppressing the outside magnetic field involves using large, immobile, and expensive magnetically shielded rooms to passively isolate the SERF OPMs from the sources of the outside magnetic field, effectively reducing the dynamic range requirements of the SERF OPMs used to measure the weak MEG signals.
These shielded rooms, however, are generally not viable for the consumer market, especially with regard to BCI applications, where it desirable that the MEG-based neural activity measurement system be incorporated into a wearable or portable form factor. Thus, for BCI applications, SERF OPMs must be capable of operating in the ambient background magnetic field of the native environment, including the Earth's magnetic field and other local sources of magnetic fields.
Another technique for suppressing the outside magnetic field without using magnetically shielded rooms involves incorporating a direct broadband feedback control system to actively null the outside magnetic field at the SERF OPM. In this case, the system actuators attempt to cancel the entire bandwidth of the outside magnetic field by applying a noise-cancelling, broadband, magnetic field to the sensors. However, such feedback control for OPM systems has not been implemented in a wearable system.
There, thus, remains a need to provide means for more effectively suppressing an outside magnetic field in a wearable neural detection system.
In accordance with a first aspect of the present inventions, a system comprises a plurality of magnetometers (e.g., a plurality of coarse magnetometers, such as flux gate magnetometers, and a plurality of fine magnetometers, such as optically pumped magnetometers (OPMs)) configured for taking measurements of an arbitrary magnetic field having one or more magnetic field components. The system further comprises a processor configured for acquiring the arbitrary magnetic field measurements from the plurality of magnetometers, and generating a generic model of at least one of the one or more magnetic field components of the arbitrary magnetic field in the vicinity of the plurality of magnetometers. The generic magnetic field model comprises an initial number of different basis functions (e.g., 0th order basis functions and 1st order basis functions or at least one non-linear basis function, such as, e.g., a vector spherical harmonics (VSH) basis function).
The processor is further configured for applying Maxwell's equations to the generic magnetic field model to reduce the initial number of different basis functions, thereby yielding a Maxwell-constrained model of the magnetic field component(s) of the arbitrary magnetic field, estimating the magnetic field component(s) of the arbitrary magnetic field at each of at least one of the plurality of magnetometers (e.g., fine magnetometers) based on the constrained magnetic field model and the arbitrary magnetic field measurements acquired from each of the magnetometer(s).
In one embodiment, the magnetic field component(s) of the magnetic field measurement acquired from each of the magnetometer(s) comprises a physical portion and a non-physical portion, and the magnetic field component estimate(s) at the magnetometer(s) has a physical portion and a non-physical portion. The non-physical portion of the magnetic field component estimate(s) at each of the magnetometer(s) is respectively less than the non-physical portion of the magnetic field component of the magnetic field measurement acquired from each of the magnetometer(s).
In another embodiment, the processor is configured for estimating the magnetic field component(s) at each of the magnetometer(s) by parameterizing the constrained magnetic field model at least partially based on the arbitrary magnetic field measurements acquired from the plurality of magnetometers, thereby yielding a parameterized model of magnetic field component(s) of the arbitrary magnetic field in the vicinity of the plurality of magnetometers, and substituting each location of the magnetometer(s) into the parameterized magnetic field model. In this embodiment, the processor may be configured for parameterizing the constrained magnetic field model by fitting the coefficients of the reduced number of basis functions of the constrained magnetic field model at least partially to the arbitrary magnetic field measurements acquired from the plurality of magnetometers (e.g., using a least squares optimization technique), and incorporating the fitted coefficients into the constrained magnetic field model.
In still another embodiment, the magnetic field component(s) of the arbitrary magnetic field of which the measurements are taken comprises an outside magnetic field and a magnetoencephalography (MEG) magnetic field, the magnetic field component(s) of the arbitrary magnetic field of which the generic model is generated comprises the outside magnetic field, the initial number of different basis functions in the generic magnetic field model comprises basis functions for the outside magnetic field, and the magnetic field component estimate(s) at each of the magnetometer(s)(s) comprises an outside magnetic field estimate.
In this embodiment, the magnetic field component(s) of the arbitrary magnetic field of which the generic model is generated may further comprise the MEG magnetic field, the initial number of different basis functions in the generic magnetic field model may further comprise basis functions for the MEG magnetic field, and the magnetic field component estimate(s) at each of the magnetometer(s) may further comprise a MEG magnetic field estimate.
In this embodiment, the arbitrary magnetic field may be a total residual magnetic field, and the system may further comprise at least one magnetic field actuator (e.g., three orthogonal magnetic field actuators, each of which may be uniform) configured for generating an actuated magnetic field that at least partially cancels the outside magnetic field at each of the magnetometer(s), thereby yielding the total residual magnetic field at each of the magnetometer(s), such that the arbitrary magnetic field measurements acquired from the plurality of magnetometers are total residual magnetic field measurements acquired from the plurality of magnetometers. In this case, the processor is configured for estimating the total residual magnetic field at each of the magnetometer(s) based on the outside magnetic field estimate at each of the magnetometer(s) and the total residual magnetic field measurements acquired from the plurality of magnetometers, and controlling the actuated magnetic field at least partially based on the total residual magnetic field estimate at each of the magnetometer(s) in a manner that suppresses the total residual magnetic field at each of the magnetometer(s) to a baseline level, such that each at each of the magnetometer(s) is in-range.
In this embodiment, the processor may be configured for estimating the total residual magnetic field at each of the magnetometer(s) by determining a known actuated magnetic field at each of the magnetometer(s) (e.g., by summing the known actuated magnetic field at each of the magnetometer(s) and the outside magnetic field estimate at each of the magnetometer(s)), and estimating the total residual magnetic field at each of the magnetometer(s) based on the known actuated magnetic field at each of the magnetometer(s) and the outside magnetic field estimate at each of the magnetometer(s). Each of the magnetic field actuator(s) may respectively have at least one actuation strength, in which case, the processor may be configured for determining the known actuated magnetic field at each of the magnetometer(s) based on a known profile of the magnetic field actuator(s) and the actuation strength(s) of the magnetic field actuator(s).
In yet another embodiment, the system further comprises a signal acquisition unit configured for being worn on a head of a user. The signal acquisition unit comprises a support structure, the magnetic field actuator(s) affixed to the support structure, and the plurality of magnetometers affixed to the support structure. The signal acquisition unit is configured for deriving a MEG signal from the total residual magnetic field estimate at each of the magnetometer(s). The system further comprises a signal processing unit configured for determining an existence of neural activity in the brain of the user at least partially based on the MEG signal derived from the total residual magnetic field estimate at each of the magnetometer(s).
In accordance with a second aspect of the present inventions, a method comprises acquiring measurements (e.g., coarse total residual magnetic field measurements and fine total residual magnetic field measurements) of an arbitrary magnetic field having one or more magnetic field components at a plurality of detection locations. The method further comprises generating a generic model of at least one of the magnetic field component(s) of the arbitrary magnetic field in the vicinity of the plurality of detection locations. The generic magnetic field model comprises an initial number of different basis functions (e.g., 0th order basis functions and 1st order basis functions or at least one non-linear basis function, such as, e.g., a vector spherical harmonics (VSH) basis function). The method further comprises applying Maxwell's equations to the generic magnetic field model to reduce the initial number of different basis functions, thereby yielding a Maxwell-constrained model of the magnetic field component(s) of the arbitrary magnetic field, estimating the at least one magnetic field component of the arbitrary magnetic field at each of at least one of the plurality of detection locations (e.g., fine detection locations) based on the constrained magnetic field model and the arbitrary magnetic field measurements acquired from each of the detection location(s).
In one method, the magnetic field component(s) of the magnetic field measurement acquired from each of the detection location(s) comprises a physical portion and a non-physical portion, and the magnetic field component estimate(s) at the detection location(s) has a physical portion and a non-physical portion. The non-physical portion of magnetic field component estimate(s) at each of the detection location(s) is respectively less than the non-physical portion of the magnetic field component(s) of the magnetic field measurement acquired from each of the detection location(s).
In another method, estimating the magnetic field component(s) at each of the detection location(s) comprises parameterizing the constrained magnetic field model at least partially based on the arbitrary magnetic field measurements acquired from the plurality of detection locations, thereby yielding a parameterized model of the magnetic field component(s) of the arbitrary magnetic field in the vicinity of the plurality of detection locations, and substituting each of the detection location(s) into the parameterized magnetic field model. In this method, parameterizing the constrained magnetic field model may comprise fitting the coefficients of the reduced number of basis functions of the constrained magnetic field model at least partially to the arbitrary magnetic field measurements acquired from the plurality of detection locations (e.g., using a least squares optimization technique), and incorporating the fitted coefficients into the constrained magnetic field model.
In still another method, the magnetic field component(s) of the arbitrary magnetic field of which the measurements are taken comprises an outside magnetic field and a magnetoencephalography (MEG) magnetic field, the magnetic field component(s) of the arbitrary magnetic field of which the generic model is generated comprises the outside magnetic field, the initial number of different basis functions in the generic magnetic field model comprises basis functions for the outside magnetic field, and the magnetic field component estimate(s) at each of the detection location(s) comprises an outside magnetic field estimate.
In this method, the magnetic field component(s) of the arbitrary magnetic field of which the generic model is generated may further comprise the MEG magnetic field, the initial number of different basis functions in the generic magnetic field model may further comprise basis functions for the MEG magnetic field, and the magnetic field component estimate(s) at each of the detection location(s) may further comprise a MEG magnetic field estimate.
In this method, the arbitrary magnetic field may be a total residual magnetic field, and the method may further comprise generating an actuated magnetic field (e.g., a uniform actuated magnetic field generated in three dimensions) that at least partially cancels the outside magnetic field at each of the magnetometer(s), thereby yielding the total residual magnetic field at each of the detection location(s), such that the arbitrary magnetic field measurements acquired from the plurality of detection locations are total residual magnetic field measurements acquired from the plurality of detection locations. In this case, the total residual magnetic field is estimated at each of the detection location(s) based on the outside magnetic field estimate at each of the detection location(s) and the total residual magnetic field measurements acquired from the plurality of detection locations, and controlling the actuated magnetic field at least partially based on the total residual magnetic field estimate at each of the detection location(s) in a manner that suppresses the total residual magnetic field at each of the detection location(s) to a baseline level, such that an accuracy of the total residual magnetic field at each of the detection location(s) increases.
In this method, total residual magnetic field at each of the detection location(s) may be estimated by determining a known actuated magnetic field at each of the detection location(s) (e.g., by summing the known actuated magnetic field at each of the detection location(s) and the outside magnetic field estimate at each of the detection location(s)), and estimating the total residual magnetic field at each of the detection location(s) based on the known actuated magnetic field at each of the detection location(s) and the outside magnetic field estimate at each of the detection location(s). The known actuated magnetic field at each of detection location may be determined based on a known profile of actuated magnetic field and an actuation strength of the actuated magnetic field.
Yet another method further comprises deriving a MEG signal from the total residual magnetic field estimate at each of the detection location(s), and determining an existence of neural activity in the brain of a user at least partially based on the MEG signal derived from the total residual magnetic field estimate at each of the detection location(s).
In accordance with a third aspect of the present inventions, a system comprises a plurality of magnetometers configured for taking measurements of a magnetic field containing a magnetoencephalography (MEG) magnetic field emanating from a brain of a user, such that the magnetic field measurement taken at each of at least one of the plurality of magnetometers has a MEG magnetic field component. The MEG magnetic field component of the magnetic field measurement taken at each of the magnetometer(s) has a physical portion and a non-physical portion.
The system further comprises a processor configured for acquiring the magnetic field measurements from the plurality of magnetometers, and suppressing the non-physical portion of the MEG magnetic field component of the magnetic field measurement acquired from each of the magnetometer(s) relative to the physical portion of the MEG magnetic field component of the magnetic field measurement acquired from each of the magnetometer(s).
In one embodiment, the magnetic field further comprises an outside magnetic field, such that the magnetic field measurement acquired from each of the magnetometer(s) further has an outside magnetic field component. In this case, the processor is configured for suppressing the outside magnetic field component of the magnetic field measurement acquired from each of the magnetometer(s) relative to the MEG magnetic field component of the magnetic field measurement acquired from each of the magnetometer(s).
In one specific implementation of this embodiment, the processor may be configured for suppressing the outside magnetic field measurement component of the magnetic field measurement acquired from each of the magnetometer(s) relative to the MEG magnetic field component of the magnetic field measurement acquired from each of the magnetometer(s) based on one or more of a temporal frequency of the outside magnetic field (e.g., by suppressing the magnetic field measurement acquired from the each at least one magnetometer at DC and harmonic temporal frequencies), a spatial frequency of the outside magnetic field (e.g., by suppressing the magnetic field measurement acquired from the each at least one magnetometer at relatively low spatial frequencies), and a strength of the outside magnetic field (e.g., by suppressing the magnetic field measurement acquired from each of the magnetometer(s) at relatively high strength frequency components).
In another embodiment, the processor is configured for suppressing the outside magnetic field measurement component of the magnetic field measurement acquired from each of the magnetometer(s) relative to the MEG magnetic field component of the magnetic field measurement acquired from each of the magnetometer(s) by generating a generic model of the MEG magnetic field in the vicinity of the plurality of magnetometers. The generic MEG magnetic field model comprises an initial number of different basis functions. The generic magnetic field model comprises an initial number of different basis functions (e.g., 0th order basis functions and 1st order basis functions or at least one non-linear basis function, such as, e.g., a vector spherical harmonics (VSH) basis function). The processor is further configured for applying Maxwell's equations to the generic MEG magnetic field model to reduce the initial number of different basis functions, thereby yielding a Maxwell-constrained model of the MEG magnetic field model, and estimating the MEG magnetic field model at each magnetometer based on the constrained MEG magnetic field model and the magnetic field measurements acquired from the plurality of magnetometers.
In this embodiment, the processor is configured for estimating the MEG magnetic field model at each of the magnetometer(s) by parameterizing the constrained MEG magnetic field model at least partially based on the magnetic field measurements acquired from the plurality of magnetometers, thereby yielding a parameterized model of the outside magnetic field in the vicinity of the plurality of magnetometers, and substituting a location of each of the magnetometer(s) into the parameterized magnetic field model. In this embodiment, the processor may be configured for parameterizing the constrained magnetic field model by fitting the coefficients of the reduced number of basis functions of the constrained magnetic field model at least partially to the arbitrary magnetic field measurements acquired from the plurality of magnetometers (e.g., using a least squares optimization technique), and incorporating the fitted coefficients into the constrained magnetic field model, e.g., the Maxwell-constrained outside magnetic field model.
In yet another embodiment, the system further comprises a signal acquisition unit configured for being worn on a head of a user. The signal acquisition unit comprises a support structure, the magnetic field actuator(s) affixed to the support structure, and the plurality of magnetometers affixed to the support structure. The signal acquisition unit is configured for deriving a MEG signal from the magnetic field estimate at each of the magnetometer(s). The system further comprises a signal processing unit configured for determining an existence of neural activity in the brain of the user at least partially based on the MEG signal derived from the magnetic field measurement at each of the magnetometer(s).
In accordance with a fourth aspect of the present inventions, a method comprises acquiring measurements of an arbitrary magnetic field respectively at a plurality of detection locations. The arbitrary magnetic field comprises a magnetoencephalography (MEG) magnetic field emanating from a brain of a user, such that the magnetic field measurement taken at each of at least one of the plurality of detection locations has a MEG magnetic field component. The MEG magnetic field component of the magnetic field measurement acquired from each detection location has a physical portion and a non-physical portion. The method further comprises suppressing the non-physical portion of the MEG magnetic field component of the magnetic field measurement acquired from each detection location relative to the physical portion of the MEG magnetic field component of the magnetic field measurement acquired from each detection location.
In one method, the magnetic field further comprises an outside magnetic field, such that the magnetic field measurement acquired from each detection location further has an outside magnetic field component, in which case, the processor is configured for suppressing the outside magnetic field component of the magnetic field measurement acquired from each detection location relative to the MEG magnetic field component of the magnetic field measurement acquired from each detection location. The outside magnetic field measurement component of the magnetic field measurement acquired from each detection location may be suppressed relative to the MEG magnetic field component of the magnetic field measurement acquired from each detection location based on one or more of a temporal frequency of the outside magnetic field (e.g., by suppressing the magnetic field measurement acquired from the each at least one magnetometer at DC and harmonic temporal frequencies), a spatial frequency of the outside magnetic field (e.g., by suppressing the magnetic field measurement acquired from the each at least one magnetometer at relatively low spatial frequencies), and a strength of the outside magnetic field (e.g., by suppressing the magnetic field measurement acquired from each of the magnetometer(s) at relatively high strength frequency components).
In another method, suppressing the outside magnetic field measurement component of the magnetic field measurement acquired from each one detection location relative to the MEG magnetic field component of the magnetic field measurement acquired from each detection location comprises generating a generic model of the MEG magnetic field in the vicinity of the plurality of detection locations.
The generic magnetic field model comprises an initial number of different basis functions (e.g., 0th order basis functions and 1st order basis functions or at least one non-linear basis function, such as, e.g., a vector spherical harmonics (VSH) basis function). The method further comprises applying Maxwell's equations to the generic magnetic field model to reduce the initial number of different basis functions, thereby yielding a Maxwell-constrained model of the MEG magnetic field model, and estimating the MEG magnetic field model at each detection location based on the constrained MEG magnetic field model and the magnetic field measurements acquired from the plurality of detection locations.
In this method, estimating the MEG magnetic field model at the each of at least one detection location comprises parameterizing the constrained MEG magnetic field model at least partially based on the magnetic field measurements acquired from the plurality of detection locations, thereby yielding a parameterized model of the outside magnetic field in the vicinity of the plurality of detection locations, and substituting each detection location into the parameterized magnetic field model. Parameterizing the constrained MEG magnetic field model may comprise fitting the coefficients of the reduced number of basis functions of the constrained magnetic field model at least partially to the magnetic field measurements acquired from the plurality of detection locations (e.g., using a least squares optimization technique), and incorporating the fitted coefficients into the Maxwell-constrained outside magnetic field model.
Yet another method comprises deriving a MEG signal from the magnetic field measurement at each detection location, and determining an existence of neural activity in the brain of the user at least partially based on the MEG signal derived from the magnetic field measurement acquired from each detection location.
In accordance with a fifth aspect of the present inventions, a system comprises a plurality of magnetometers (e.g., a plurality of coarse magnetometers, such as flux gate magnetometers, and a plurality of fine magnetometers, such as optically pumped magnetometers (OPMs)) configured for taking measurements of an arbitrary magnetic field having a plurality of magnetic field components. The system further comprises a processor configured for acquiring the arbitrary magnetic field measurements from the plurality of magnetometers, and generating a generic model of the plurality of magnetic field components of the arbitrary magnetic field in the vicinity of the plurality of magnetometers. The generic magnetic field model comprises a plurality of basis functions having multiple sets of basis functions respectively corresponding to the plurality of magnetic field components of the arbitrary magnetic field, and the processor is further configured for parameterizing the generic magnetic field model by simultaneously fitting coefficients of the plurality of basis functions at least partially to the arbitrary magnetic field measurements acquired from the plurality of magnetometers (e.g., using a least squares optimization technique), thereby yielding a parameterized model of the plurality of magnetic field components of the arbitrary magnetic field in the vicinity of the plurality of magnetometers.
In one embodiment, the plurality of magnetic field components of the arbitrary magnetic field comprises a physical portion of an outside magnetic field and a non-physical portion of the outside magnetic field, the first set of basis functions correspond to modes of the outside magnetic field that are physically possible, and the second set of basis functions correspond to modes of the outside magnetic field that are physically impossible. In another embodiment, the plurality of magnetic field components of the arbitrary magnetic field comprises a magnetoencephalography (MEG) magnetic field and an outside magnetic field, the first set of basis functions correspond to modes in the MEG magnetic field, and the second set of basis functions correspond to modes in the outside magnetic field. In still another embodiment, the plurality of magnetic field components of the arbitrary magnetic field comprises a magnetoencephalography (MEG) magnetic field of interest and a magnetoencephalography (MEG) magnetic field not of interest, the first set of basis functions correspond to modes of the MEG magnetic field of interest, and the second set of basis functions correspond to modes of the MEG magnetic field not of interest.
The processor is further configured for estimating a first one of the plurality of magnetic field components of the arbitrary magnetic field at each of at least one of the plurality of magnetometers (e.g., fine magnetometers) based on a first one of the multiple sets of basis functions of the parameterized magnetic field model. The processor may be configured for estimating the first one of the plurality of magnetic field components of the arbitrary magnetic field at each of the magnetometer(s) based on the parameterized magnetic field model by substituting a location of each of the magnetometer(s) into the parameterized magnetic field model.
In one embodiment, the processor is further configured for estimating a second one of the plurality of magnetic field components of the arbitrary magnetic field at each of the magnetometer(s) based on a second one of the multiple sets of basis functions of the parameterized magnetic field model.
In another embodiment, the generic magnetic field model comprises a coefficient vector and a matrix of influence from the coefficient vector to the plurality of magnetic field components of the arbitrary magnetic field. The coefficient vector has a p number of coefficients respectively corresponding to the plurality of basis functions. The influence matrix comprises a p number of column vectors and an N number of row vectors respectively corresponding to the arbitrary magnetic field measurements acquired from the plurality of magnetometers, where p is less than N. The processor is configured for simultaneously fitting the coefficients of the plurality of basis functions at least partially to the arbitrary magnetic field measurements acquired from the plurality of magnetometers by equating the product of the coefficient vector and the influence matrix to the arbitrary magnetic field measurements acquired from the plurality of magnetometers, and simultaneously fitting the p number of coefficients in the coefficient vector at least partially to the arbitrary magnetic field measurements acquired from the plurality of magnetometers.
In still another embodiment, the plurality of magnetic field components of the arbitrary magnetic field comprises an outside magnetic field, and the estimated first one of the plurality of magnetic field components at each of the magnetometer(s) is an outside magnetic field estimate at each of the magnetometer(s). In this embodiment, the system further comprises at least one magnetic field actuator (e.g., three orthogonal magnetic field actuators, each of which may be uniform) configured for generating an actuated magnetic field that at least partially cancels the outside magnetic field at each of the magnetometer(s), thereby yielding a total residual magnetic field at each of the magnetometer(s) as the arbitrary magnetic field. In this embodiment, the processor is configured for estimating the total residual magnetic field at each of the magnetometer(s) based on the outside magnetic field estimate at each of the magnetometer(s) and the total residual magnetic field measurements acquired from the plurality of magnetometers.
In one specific implementation of this embodiment, the processor may be configured for estimating the total residual magnetic field at each of the magnetometer(s) by determining a known actuated magnetic field at each of the magnetometer(s) (e.g., by summing the known actuated magnetic field at each of the magnetometer(s) and the outside magnetic field estimate at each of the magnetometer(s)), and estimating the total residual magnetic field at each of the magnetometer(s) based on the known actuated magnetic field at each of the magnetometer(s) and the outside magnetic field estimate at each of the magnetometer(s). Each of the magnetic field actuator(s) may respectively have at least one actuation strength, in which case, the processor may be configured for determining the known actuated magnetic field at each of the magnetometer(s) based on a known profile of the magnetic field actuator(s) and the actuation strength(s) of the magnetic field actuator(s).
In this embodiment, the processor is further configured for controlling the actuated magnetic field at least partially based on the total residual magnetic field estimate at each of the magnetometer(s) in a manner that suppresses the total residual magnetic field at each of the magnetometer(s) to a baseline level, such that each of the magnetometer(s) is in-range.
In this embodiment, the system may further comprise a signal acquisition unit configured for being worn on a head of a user. The signal acquisition unit comprises a support structure, the magnetic field actuator(s) affixed to the support structure, and the plurality of magnetometers affixed to the support structure. The signal acquisition unit is configured for deriving at least one MEG signal(s) from the total residual magnetic field estimate at each of the magnetometer(s). The system further comprises a signal processing unit configured for determining an existence of neural activity in the brain of the user at least partially based on the MEG signal(s) derived from the total residual magnetic field estimate at each of the magnetometer(s).
In accordance with a sixth aspect of the present inventions, a method comprises acquiring measurements (e.g., coarse total residual magnetic field measurements and fine total residual magnetic field measurements) of an arbitrary magnetic field having a plurality of magnetic field components respectively from a plurality of detection locations. The method further comprises generating a generic model of the plurality of magnetic field components of the arbitrary magnetic field in the vicinity of the plurality of detection locations. The generic magnetic field model comprises an initial plurality of basis functions having multiple sets of basis functions.
The method further comprises parameterizing the generic magnetic field model by simultaneously fitting coefficients of the plurality of basis functions at least partially to the arbitrary magnetic field measurements acquired from the plurality of detection locations (e.g., using a least squares optimization technique), thereby yielding a parameterized model of the plurality of magnetic field components of the arbitrary magnetic field in the vicinity of the plurality of detection locations.
In one method, the plurality of magnetic field components of the arbitrary magnetic field comprises a physical portion of an outside magnetic field and a non-physical portion of the outside magnetic field, the first set of basis functions correspond to modes of the outside magnetic field that are physically possible, and the second set of basis functions correspond to modes of the outside magnetic field that are physically impossible. In another method, the plurality of magnetic field components of the arbitrary magnetic field comprises a magnetoencephalography (MEG) magnetic field and an outside magnetic field, the first set of basis functions correspond to modes in the MEG magnetic field, and the second set of basis functions correspond to modes in the outside magnetic field. In still another method, the plurality of magnetic field components of the arbitrary magnetic field comprises a magnetoencephalography (MEG) magnetic field of interest and a magnetoencephalography (MEG) magnetic field not of interest, the first set of basis functions correspond to modes of the MEG magnetic field of interest, and the second set of basis functions correspond to modes of the MEG magnetic field not of interest.
The method further comprises estimating a first one of the plurality of magnetic field components of the arbitrary magnetic field at each of at least one of the plurality of detection locations (e.g., fine detection locations) based on a first one of the multiple sets of basis functions. Estimating the first one of the plurality of magnetic field components of the arbitrary magnetic field at each of the detection location(s) based on the parameterized magnetic field model may comprise substituting each detection location(s) into the parameterized magnetic field model.
One method further comprises estimating a second one of the plurality of magnetic field components of the arbitrary magnetic field at each of the detection location(s) based on a second one of the multiple sets of basis functions.
In another method, the generic magnetic field model comprises a coefficient vector and a matrix of influence from the coefficient vector to the plurality of magnetic field components of the arbitrary magnetic field, the coefficient vector having a p number of coefficients respectively corresponding to the plurality of basis functions. The influence matrix comprises a p number of column vectors and an N number of row vectors respectively corresponding to the arbitrary magnetic field measurements acquired from the plurality of detection locations, where p is less than N. The coefficients of the plurality of basis functions may be simultaneously fitted at least partially to the arbitrary magnetic field measurements acquired from the plurality of detection locations by equating the product of the coefficient vector and the influence matrix to the arbitrary magnetic field measurements acquired from the plurality of detection locations and simultaneously fitting the p number of coefficients in the coefficient vector at least partially to the arbitrary magnetic field measurements acquired from the plurality of detection locations.
In still another method, the magnetic field component(s) of the arbitrary magnetic field comprises an outside magnetic field, and the estimated first one of the plurality of magnetic field components at each of the detection location(s) is an outside magnetic field estimate at each of the detection location(s). This method further comprises generating an actuated magnetic field (e.g., a uniform actuated magnetic field generated in three dimensions) that at least partially cancels an outside magnetic field at each of the detection location(s), thereby yielding a total residual magnetic field as the arbitrary magnetic field at each of the detection location(s). This method further comprises estimating the total residual magnetic field at each of the detection location(s) based on the outside magnetic field estimate at each of the detection location(s) and the total residual magnetic field measurements acquired from the plurality of detection locations (e.g., by summing the known actuated magnetic field at each of the detection location(s) and the outside magnetic field estimate at each of the detection location(s)). In this method, the known actuated magnetic field may be determined at each of the detection location(s) based on a known profile of the actuated magnetic field an actuation strength of the actuated magnetic field.
The method further comprises controlling the actuated magnetic field at least partially based on the total residual magnetic field estimate at each of the detection location(s) in a manner that suppresses the total residual magnetic field at each of the detection location(s) to a baseline level, such that an accuracy of the total residual magnetic field measurement acquired from each of the detection location(s) increases.
This method may further comprise deriving a MEG signal from the total residual magnetic field estimate at each of the detection location(s), and determining an existence of neural activity in the brain of a user at least partially based on the MEG signal derived from the total residual magnetic field estimate at each of the detection location(s).
Other and further aspects and features of the invention will be evident from reading the following detailed description of the preferred embodiments, which are intended to illustrate, not limit, the invention.
The drawings illustrate the design and utility of preferred embodiments of the present invention, in which similar elements are referred to by common reference numerals. In order to better appreciate how the above-recited and other advantages and objects of the present inventions are obtained, a more particular description of the present inventions briefly described above will be rendered by reference to specific embodiments thereof, which are illustrated in the accompanying drawings.
Understanding that these drawings depict only typical embodiments of the present inventions and are not therefore to be considered limiting of its scope, the present inventions will be described and explained with additional specificity and detail through the use of the accompanying drawings in which:
Significantly, the neural activity measurement systems (and variations thereof) described herein are configured for non-invasively acquiring magnetoencephalography (MEG) signals from a brain of a user while effectively cancelling an outside magnetic field without the use of magnetically shielded rooms, and identifying and localizing the neural activity within the cortical structures of the brain of the user based on the acquired magnetoencephalography (MEG) signals.
The neural activity measurement system described herein may take the form of a brain computer interface (BCI) (also known as a neural-controlled interface (NCI), mind-machine interface (MMI), direct neural interface (DNI), or brain-machine interface (BMI)), which converts the neural activity information into commands that are output to an external device or devices for carrying out desired actions that replace, restore, enhance, supplement, or improve natural central nervous system (CNS) output, and thereby changes the ongoing interactions between the CNS of a user and an external or internal environment.
For example, as illustrated in
To this end, the neural activity measurement system 10 generally comprises a signal acquisition unit 18 configured for at least partially cancelling a relatively strong outside magnetic field BOUT within an environmental magnetic field BENV that also includes a relatively weak MEG magnetic field BMEG induced by electrical current (indicative of neural activity) in a brain 14 of a user 12. That is, BTOT=BENV+BACT=BOUT+BMEG+BACT. The outside magnetic field BOUT may emanate from global sources (e.g., the Earth's magnetic field), and from localized sources, including, but not limited to, from electromagnetic radiation emanating from electrical outlets and sockets, electrical wires or connections in the wall, and everyday electrical equipment (microwave ovens, televisions, refrigerators, environmental systems (air conditioning, etc.) in a home, office, or laboratory setting, as well as from cell phones, biomagnetics unrelated to neural signals (such as facial muscles, magnetic fields produced by the heart or nerves firing), everyday objects encountered inside (metal and magnetic objects, including steel supports, rebar, studs, utility boxes, etc.) and outside spaces, such as cell phone towers, power lines, transformers, and moving vehicles (e.g., cars, trains, bikes, electric bikes and scooters, electric cars, etc.), user motion/rotation/translation in a background field (earth field), user clothing and eyeglasses, personal electronics (e.g., laptop computers, watches, phones, smart rings, etc.), active implantable medical devices (pacemakers), augmented reality/virtual reality, sound systems (that use magnets), etc.
The signal acquisition unit 18 is configured for generating an actuated magnetic field BACT that at least partially cancels the relative strong outside magnetic field BOUT within the environmental magnetic field BENV, yielding a total residual magnetic field BTOT (which is preferably zero or near-zero due to the summation of the environmental magnetic field BENV and the actuated magnetic field BACT). The signal acquisition unit 18 is further configured for detecting the total residual magnetic field BTOT as feedback to cancel the outside magnetic field BOUT. The signal acquisition unit 18 is also configured for extracting and outputting a clean (i.e., reduced-noise) electrical MEG signals SMEG of the MEG magnetic field BMEG from the total residual magnetic field BTOT.
The signal acquisition unit 18 may utilize any suitable technique for acquiring the MEG magnetic field BMEG, including, but not limited to the techniques described in U.S. patent application Ser. No. 16,428,871, entitled “Magnetic Field Measurement Systems and Methods of Making and Using,” U.S. patent application Ser. No. 16/418,478, entitled “Magnetic Field Measurement System and Method of Using Variable Dynamic Range Optical Magnetometers”, U.S. patent application Ser. No. 16/418,500, entitled, “Integrated Gas Cell and Optical Components for Atomic Magnetometry and Methods for Making and Using,” U.S. patent application Ser. No. 16/457,655, entitled “Magnetic Field Shaping Components for Magnetic Field Measurement Systems and Methods for Making and Using,” U.S. patent application Ser. No. 16/213,980, entitled “Systems and Methods Including Multi-Basis function Operation of Optically Pumped Magnetometer(s),” (now U.S. Pat. No. 10,627,460), U.S. patent application Ser. No. 16/456,975, entitled “Dynamic Magnetic Shielding and Beamforming Using Ferrofluid for Compact Magnetoencephalography (MEG),” U.S. patent application Ser. No. 16/752,393, entitled “Neural Feedback Loop Filters for Enhanced Dynamic Range Magnetoencephalography (MEG) Systems and Methods,” U.S. patent application Ser. No. 16/741,593, entitled “Magnetic Field Measurement System with Amplitude-Selective Magnetic Shield,” U.S. Provisional Application Ser. No. 62/858,636, entitled “Integrated Magnetometer Arrays for Magnetoencephalography (MEG) Detection Systems and Methods,” U.S. Provisional Application Ser. No. 62/836,421, entitled “Systems and Methods for Suppression of Non-Neural Interferences in Magnetoencephalography (MEG) Measurements,” U.S. Provisional Application Ser. No. 62/842,818 entitled “Active Shield Arrays for Magnetoencephalography (MEG),” U.S. Provisional Application Ser. No. 62/926,032 entitled “Systems and Methods for Multiplexed or Interleaved Operation of Magnetometers,” U.S. Provisional Application Ser. No. 62/896,929 entitled “Systems and Methods having an Optical Magnetometer Array with Beam Splitters,” and U.S. Provisional Application Ser. No. 62/960,548 entitled “Methods and Systems for Fast Field Zeroing for Magnetoencephalography (MEG),” which are all expressly incorporated herein by reference.
The neural activity measurement system 10 further comprises a signal processing unit 20 configured for processing the electrical MEG signal SMEG to identify and localize neural activity within the cortex of the brain 14 of the user 12, and issuing the commands CMD to the external device 16 in response to the identified and localized neural activity in the brain 14 of the user 12.
It should be appreciated that, although the neural activity measurement system 10 is described herein in the context of a BCI, the present inventions should not be so limited, and may be applied to any system used for any application (including, but not limited to, medical, entertainment, neuromodulation stimulation, lie detection devices, alarm, educational, etc.), where it is desirable to perform measurements on a magnetic field induced by any physiological process in a person that would benefit from cancelling the outside magnetic field BOUT. For example, instead of deriving neural activity information from MEG signals, magnetic fields induced by electrical heart activity can be measured to determine heart activity information of a person.
Furthermore, it should also be appreciated that, although the use of the signal acquisition unit lends itself well to neural activity measurement systems, the signal acquisition unit 18 may find use in other applications, such as, e.g., other types of biomedical sensing, vehicle navigation, mineral exploration, non-destructive testing, detection of underground devices, asteroid mining, space exploration, etc. Thus, signal acquisition unit 18 can be adapted to measure neural signals generated from non-brain anatomical structures, as well as other types of biological signals and non-biological signals.
Referring now to
As shown, the signal acquisition unit 18 is configured for being applied to the user 12, and in this case, worn on the head of the user 12. The signal acquisition unit 18 comprises a support structure 24, a plurality of magnetometers 26 (divided between a plurality of coarse magnetometers 26a and a plurality of fine magnetometers 26b) distributed about the support structure 24, a set of magnetic field actuators 28 in proximity to the fine magnetometers 26b, and a processor 30 electrically coupled between the magnetometers 26 and the set of actuators 28.
The support structure 24 may be shaped, e.g., have a banana, headband, cap, helmet, beanie, other hat shape, or other shape adjustable and conformable to the user's head, such that at least some of the magnetometers 26 are in close proximity, preferably in contact, with the outer skin of the head, and in this case, the scalp of the user 12. The support structure 24 may be made out of any suitable cloth, soft polymer, plastic, hard shell, and/or any other suitable material as may serve a particular implementation. An adhesive, strap, or belt (not shown) can be used to secure the support structure 24 to the head of the user 12.
Each of the magnetometers 26 is configured for detecting a spatial component of the total residual magnetic field BTOT, and outputting a corresponding electrical signal representative of the spatial component of the total residual magnetic field BTOT. In the illustrated embodiment, the plurality of coarse magnetometers 26a is distributed on the outside of the support structure 24 for detecting the respective spatial components of the total residual magnetic field BTOT mainly from outside of the support structure 24, whereas the plurality of fine magnetometers 26b is distributed on the inside of the support structure 24 for detecting the respective spatial components of the total residual magnetic field BTOT mainly from inside the support structure 24 (i.e. they are closer to the brain 14 of the user 12).
Each of the coarse magnetometers 26a has a relatively low sensitivity, but high dynamic sensitivity range, to magnetic fields, whereas each of the fine magnetometers 26b has a relatively high sensitivity, but low dynamic sensitivity range. The signal acquisition unit 18 may have any suitable number of magnetometers 26. For example, the signal acquisition unit 18 may have twelve coarse magnetometers 26a and twenty-five fine magnetometers 26b, although one of ordinary skill in the art would understand that signal acquisition unit 18 may have any suitable number of coarse magnetometers 26a and magnetometers 26b, including more coarse magnetometers 26a then fine magnetometers 26b. In alternative embodiments of the signal acquisition unit 18, the plurality of magnetometers 26 may only comprise a plurality of fine magnetometers 26b distributed on the inside of the support structure 24.
In the illustrated embodiment, each coarse magnetometer 26a takes the form of a flux gate magnetometer, which has a relatively low sensitivity (e.g., on the order of 100 fT), and thus, may not be capable of measuring weak magnetic fields generated by neural activity in the brain 14 of the user 12. However, a flux gate magnetometer has a relatively high dynamic sensitivity range (in the range of 100 fT to close to 100 μT), and thus, may operate in a large outside magnetic field BOUT. Although each of the coarse magnetometers 26a are described as taking the form of a flux gate magnetometer, other types of coarse magnetometers can be used, including, but not limited to, anisotropic magnetoresistance (AMR) sensors, tunnel magnetoresistance (TMR) sensors, Hall-effect sensors, nitrogen vacancy sensors, or any other magnetometer that can operate in a linear range over the amplitude range of a typical outside magnetic field BOUT. As will be described in further detail below, each of the coarse magnetometers 26a is specifically designed to facilitate the calibration of its offset and gain using novel pre-calibration and dynamic calibration techniques.
In the illustrated embodiment, each fine magnetometer 26b takes the form of a Spin Exchange Relaxation Free (SERF) Optically Pumped Magnetometer (OPM). Although a SERF OPM has a relatively small dynamic range (e.g., in the range of 1 ft to 200 nT), it has a relatively high sensitivity (on the order of 1 fT) to magnetic fields compared to flux gate magnetometers. Further details of SERF OPMs are described in U.S. Provisional Application Ser. No. 62/975,693, entitled “Nested and Parallel Feedback Control Loops For Ultra-Fine Measurements of Magnetic Fields From the Brain Using a Wearable MEG System” (Attorney Docket No. KERN-079), which is expressly incorporated herein by reference.
In the illustrated embodiment, each of the coarse magnetometers 26a and fine magnetometers 26b are vector magnetometers that are capable of detecting the total residual magnetic field BTOT in three dimensions (x, y, and z). For example, each of the coarse magnetometers 26a may include a triad of the scalar magnetometers, as described in U.S. Provisional Application Ser. No. 62/975,709, entitled “Self-Calibration of Flux Gate Offset and Gain Drift To Improve Measurement Accuracy Of Magnetic Fields From the Brain Using a Wearable MEG System” (Attorney Docket No. KERN-078), and each of the fine magnetometer 26b may be vector magnetometers, as described in U.S. patent application Ser. No. 16/752,393, entitled “Neural Feedback Loop Filters for Enhanced Dynamic Range Magnetoencephalography (MEG) Systems and Methods,” which are expressly incorporated herein by reference.
The clean (i.e., reduced-noise) electrical MEG signals SMEG that are representative of the spatial components of the MEG magnetic field BMEG, and that will be processed by the signal processing unit 20 for determining and localizing neural activity in the brain 14 of the user 12, will be respectively derived from the electrical signals output by the respective fine magnetometers 26b, and in some cases, from the electrical signals output by the coarse magnetometers 26a; whereas the characteristics (namely amplitude and phase) of the actuated magnetic field BACT will be derived from the electrical signals output by the respective coarse magnetometers 26a and/or the electrical signals output by at least some of the respective fine magnetometers 26b.
The set of magnetic field actuators 28 is configured for generating the actuated magnetic field BACT to at least partially cancel the outside magnetic field BOUT in the vicinity of the plurality of fine magnetometers 26b. The set of magnetic field actuators 28 may, e.g., comprise at least one coil and at least one driver that drives the coil(s) with electrical current at a defined amperage, voltage, or some other variable, and at a defined frequency, thereby setting the actuation strengths of the magnetic field actuators 28. In the illustrated embodiment, the set of magnetic field actuators 28 comprises a triad of uniform magnetic field actuators 28a-28c for respectively generating x-, y-, and z-components of the actuated magnetic field BACT to cancel the outside magnetic field BOUT in all three dimensions. In an optional embodiment, the set of magnetic field actuators 28 may also comprise six gradient magnetic field actuators (not shown) for generating first-order x-, y-, and z-gradient components of the actuated magnetic field BACT. One of ordinary skill in the art would appreciate that the set of field actuators 28 may include any suitable and type of magnetic field actuators capable of cancelling the outside magnetic field BOUT at the magnetometers 26.
The processor 30 is electrically coupled between the magnetometers 26 and magnetic field actuators 28 via electrical wires (not shown), and is configured for processing the electrical signals respectively output by the coarse magnetometers 26a (and in some cases the electrical signals output by the fine magnetometers 26b) in response to the detection of the spatial components of the total residual magnetic field BTOT, determining the characteristics of the actuated magnetic field BACT required to cancel the outside magnetic field BOUT in the total residual magnetic field BTOT, and generating cancellation control signals based on this determination that are output to the set of magnetic field actuators 28. Further details discussing novel techniques for cancelling the outside magnetic field BOUT in the total residual magnetic field BTOT are described in U.S. Provisional application Ser. No. 62/xxx,xxx, entitled “Nested and Parallel Feedback Control Loops For Ultra-Fine Measurements of Magnetic Fields From the Brain Using a Wearable MEG System” (Attorney Docket No. KERN-079).
To minimize the size, weight, and cost of the signal acquisition unit 18, the functions of the processor 30 are preferably performed digitally (e.g., in firmware, such as a programmable logic device (e.g., a field programmable gate array (FPGA), or an ASIC (application specific integrated circuit) device, or in a micro-processor)), in which case, one or more analog-to-digital converters (not shown) can be employed between the magnetometers 26 and the processor 30, and one or more digital-to-analog converters (not shown) can be employed between the magnetic field actuators 28 and the processor 30. However, it should be appreciated that, in alternative embodiments, the functions of the processor 30 may be at least partially performed in an analog fashion.
It should be noted that, although the signal acquisition unit 18 is illustrated in
The signal processing unit 20 is configured for being applied to the user 12, and in this case, worn remotely from the head of the user 12, e.g., worn on the neck, shoulders, chest, or arm) of the user 12. The signal processing unit 20 comprises a housing 36 containing a processor 38 and a controller 40. The processor 38 is configured for identifying and localizing neural activity within the cortex of the brain 14 of the user 12, and the controller 40 is configured for issuing commands CMD to an external device 16 in response to the identified and localized neural activity in the brain 14 of the user 12, as well as controlling the high-level operational functions of the signal acquisition unit 18. The signal processing unit 20 may additionally include a power supply (which if head-worn, may take the form of a rechargeable or non-chargeable battery), a control panel with input/output functions, a display, and memory. Alternatively, power may be provided to the signal processing unit 20 wirelessly (e.g., by induction).
In the illustrated embodiment, the neural activity measurement system 10 further comprises a wired connection 42 (e.g., electrical wires) for providing power from the signal processing unit 20 to the signal acquisition unit 18 and communicating between the signal processing unit 20 and the signal acquisition unit 18. Alternatively, the neural activity measurement system 10 may use a non-wired connection (e.g., wireless radio frequency (RF) signals (e.g., Bluetooth, Wifi, cellular, etc.) or optical links (e.g., fiber optic or infrared (IR)) for providing power from the signal processing unit 20 to the signal acquisition unit 18 and/or communicating between the signal processing unit 20 and the signal acquisition unit 18.
In the illustrated embodiment, the neural activity measurement system 10 further comprises a wired connection 44 (e.g., electrical wires) for providing power from the signal processing unit 20 to the external device 16 and communicating between the signal processing unit 20 and the external device 16. Alternatively, the neural activity measurement system 10 may use a non-wired connection (e.g., wireless radio frequency (RF) signals (e.g., Bluetooth, Wifi, cellular, etc.) or optical links (e.g., fiber optic or infrared (IR)) for providing power from the signal processing unit 20 to the external device 16 and/or communicating between the signal processing unit 20 and the external device 16.
The neural activity measurement system 10 may optionally comprise a remote processor 22 (e.g., a Smartphone, tablet computer, or the like) in communication with the signal processing unit 20 coupled via a wired connection (e.g., electrical wires) or a non-wired connection (e.g., wireless radio frequency (RF) signals (e.g., Bluetooth, Wifi, cellular, etc.) or optical links (e.g., fiber optic or infrared (IR)) 46. The remote processor 22 may store data from previous sessions, and include a display screen.
It should be appreciated that at least a portion of the signal acquisition and magnetic field cancellation functionality of the processor 30 in the signal acquisition unit 18 may be implemented in the signal processing unit 20, and/or at least a portion of the neural activity determination and localization functionality of the signal processing unit 20 may be implemented in the signal acquisition unit 18. In the preferred embodiment, the functionalities of the processor 30 in the signal acquisition unit 18, as well as the processor 38 and a controller 40 in the signal processing unit 20, may be implemented using one or more suitable computing devices or digital processors, including, but not limited to, a microcontroller, microprocessor, digital signal processor, graphical processing unit, central processing unit, application specific integrated circuit (ASIC), field programmable gate array (FPGA), and/or programmable logic unit (PLU). Such computing device(s) or digital processors may be associated with non-transitory computer- or processor-readable medium that stores executable logic or instructions and/or data or information, which when executed, perform the functions of these components. The non-transitory computer- or processor-readable medium may be formed as one or more registers, for example of a microprocessor, FPGA, or ASIC, or can be a type of computer-readable media, namely computer-readable storage media, which may include, but is not limited to, RAM, ROM, EEPROM, flash memory, or other memory technology, CD-ROM, digital versatile disks (“DVD”) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can be accessed by a computing device.
The signal acquisition unit 18 takes advantage of the high dynamic range of the coarse magnetometers 26a to compensate for the relatively low dynamic range of the fine magnetometers 26b to cancel the large outside magnetic field BOUT, while also taking advantage of high sensitivity of the fine magnetometers 26b to compensate for the low sensitivity of the coarse magnetometers 26a to measure the MEG signal SMEG.
In particular, and with reference to
In particular, the coarse feedback control loop 50 and fine feedback control loop 52 are implemented in the processor 30, with the coarse feedback control loop 50 coarsely controlling the set of magnetic field actuators 28 in response to input from the coarse magnetometers 26a, and the fine feedback control loop 52 finely controlling the set of magnetic field actuators 28 in response to input from the fine magnetometers 26b. Although the coarse feedback control loop 50 is illustrated as receiving input from three coarse magnetometers 26a, and the fine feedback control loop 52 is illustrated as receiving input from three fine magnetometers 26b, it should be appreciated that the coarse feedback control loop 50 can receive input from more or less coarse magnetometers 26a, including only one coarse magnetometer 26a, and the fine feedback control loop 52 can receive input from more or less fine magnetometers 26b, including only one fine magnetometer 26b. Furthermore, although the coarse feedback control loop 50 and fine feedback control loop 52 are illustrated as receiving input from an equal number of coarse magnetometers 26a and fine magnetometers 26b, the coarse feedback control loop 50 and fine feedback control loop 52 may receive input from an unequal number of coarse magnetometers 26a and fine magnetometers 26b, including a number of coarse magnetometers 26a that is greater or less the number of fine magnetometers 26b.
Initially, due to the relatively low dynamic range of the fine magnetometers 26b, the magnitude of the total residual magnetic field BTOT is too great for the fine magnetometers 26b to detect the total residual magnetic field BTOT. However, due to the relatively high dynamic range of the coarse magnetometers 26a, the spatial components of the total residual magnetic field BTOT can be respectively detected by the coarse magnetometers 26a, which outputs coarse error signals SCERR corresponding to the spatial components of the detected total residual magnetic field BTOT.
When the magnitude of the total residual magnetic field BTOT is above the dynamic range of the fine magnetometers 26b, the processor 30 acquires the coarse error signals SCERR output by the coarse magnetometers 26a in response to detecting the spatial components of the total residual magnetic field BTOT, computes the characteristics (namely, the amplitude and phase) of the actuated magnetic field BACT estimated to minimize the coarse error signals SCERR output by the coarse magnetometers 26a, and generates a corresponding noise-cancelling control signal C for output to the set of magnetic field actuators 28 for at least partially cancelling the outside magnetic field BOUT at the fine magnetometers 26b, and ultimately suppressing the total residual magnetic field BTOT to a baseline level at the fine magnetometers 26b.
In one embodiment, the processor 30 may estimate the spatial components of the total residual magnetic field BTOT respectively at each fine magnetometer 26b based on the coarse error signals SCERR output by the coarse magnetometers 26a or fine error signals SFERR of other fine magnetometers 26b, e.g., using the estimation techniques described in U.S. Provisional Application Ser. No. 62/975,719, entitled “Estimating the Magnetic Field at Distances From Direct Measurements to Enable Fine Sensors to Measure the Magnetic Field from the Brain by Using a Wearable MEG System” (Attorney Docket No. KERN-080PR01), which is expressly incorporated herein by reference.
In the embodiment illustrated in
Hence, in this example, ignoring the minute contribution of the MEG magnetic field BMEG for purposes of simplicity, the coarse magnetometers 26a and fine magnetometers 26b will measure a different total residual magnetic field BTOT=BOUT+BACT, because even though the outside magnetic field BOUT may be the same at both coarse magnetometers 26a and fine magnetometers 26b, the actuated magnetic field BACT will differ between the coarse magnetometers 26a and fine magnetometers 26b based on their different proximities to the magnetic field actuators 28. Thus, absent estimation of the spatial components of the total residual magnetic field BTOT respectively at each fine magnetometer 26b, cancellation of the outside magnetic field BOUT, and the resulting suppression of the total residual magnetic field BTOT, at the fine magnetometers 26b based directly (i.e., without correction) on the coarse error signals SCERR output by the coarse magnetometers 26a may be insufficient.
In accordance with the noise-cancelling control signal C output by the processor 30, the set of magnetic field actuators 28 generates the actuated magnetic field BACT, which combines with the outside magnetic field BOUT (along with weak MEG magnetic field BMEG from the brain 14) to create a total residual magnetic field BTOT at the fine magnetometers 26b having spatial components that are at baseline level within the operating range of the fine magnetometers 26b.
Once the spatial components of the total residual magnetic field BTOT are at the baseline level, they can be respectively detected by the fine magnetometers 26b, which outputs fine error signals SFERR corresponding to the spatial components of the detected total residual magnetic field BTOT. The processor 30 then acquires the fine error signals SFERR output by the fine magnetometers 26b in response to detecting the spatial components of the total residual magnetic field BTOT, computes the characteristics of the actuated magnetic field BACT estimated to minimize the fine error signals SFERR output by the fine magnetometers 26b, and generates a corresponding noise-cancelling control signal C for output to the set of magnetic field actuators 28 for at least partially cancelling the outside magnetic field BOUT at the fine magnetometers 26b, and ultimately suppressing the total residual magnetic field BTOT to a lower level than the baseline level at the fine magnetometers 26b.
In one embodiment, even when the spatial components of the total residual magnetic field BTOT are at the baseline level, and the fine error signals SFERR output by the fine magnetometers 26b are being actively acquired, the processor 30 may be further configured for correcting or refining the fine error signals SFERR using the estimation techniques described in U.S. Provisional Application Ser. No. 62/975,719, entitled “Estimating the Magnetic Field at Distances From Direct Measurements to Enable Fine Sensors to Measure the Magnetic Field from the Brain by Using a Wearable MEG System” (Attorney Docket No. KERN-080PR01).
In accordance with the noise-cancelling control signal C output by the processor 30, the set of magnetic field actuators 28 generates the actuated magnetic field BACT, which combines with the outside magnetic field BOUT (along with weak MEG magnetic field BMEG from the brain 14) to create a total residual magnetic field BTOT having spatial components at the fine magnetometers 26b that are at the baseline level. At this point, the fine error signals SFERR can serve to collect MEG signals SMEG representative of the spatial components of the MEG magnetic field BMEG for further processing by the signal processing unit 20 to identify and localize neural activity in the brain 14 of the user 12.
It should be appreciated that, in the illustrated embodiment, the coarse magnetometers 26a and fine magnetometers 26b are capable of detecting the total residual magnetic field BTOT in three dimensions (x, y, and z), and the set of magnetic field actuators 28 includes three magnetic field actuators 28a-28c (shown in
In an alternative embodiment, the signal acquisition unit 18 (shown in
Whether the signal acquisition unit 18 employs both the coarse feedback control loop 50 and the fine feedback control loop 52 to cancel the outside magnetic field BOUT, or employs only the coarse feedback control loop 50 to cancel the outside magnetic field BOUT, it can be appreciated that the signal acquisition unit 18 is capable of coarsely canceling a large portion of the outside magnetic field BOUT, while still collecting signals from the fine magnetometers 26b sensitive enough to measure the weaker MEG magnetic field BMEG generated by the neural activity in the brain 14 of the user 12.
The processor 30 employs the management control loop 54 to manage how the coarse feedback control loop 50 and fine feedback control loop 52 are employed (e.g., how the coarse error signals SCERR output by the coarse magnetometers 26a and the fine error signals SFERR output by the fine magnetometers 26b are to be used) for optimal cancellation of the outside magnetic field BOUT, and thus, optimal suppression of the total residual magnetic field BTOT, and corrects additional factors that can change more slowly over time, such as, e.g., calibrating the magnetometers 26 (e.g., using calibration techniques described in U.S. Provisional Application Ser. No. 62/975,709, entitled “Self-Calibration of Flux Gate Offset and Gain Drift To Improve Measurement Accuracy Of Magnetic Fields From the Brain Using a Wearable MEG System” (Attorney Docket No. KERN-078), which is expressly incorporated herein by reference), and optimizing performance metrics in the signal acquisition unit 18, either globally or locally (e.g., using optimal control methods disclosed in U.S. Provisional Application Ser. No. 62/975,727, entitled “Optimal Methods to Feedback Control and Estimate Magnetic Fields to Enable a Wearable MEG System to Measure Magnetic Fields from the Brain” (Attorney Docket No. KERN-082), which is expressly incorporated herein by reference), adapting to changing time delays in computations, etc. Further details discussing the functioning of the management control loop 54 are disclosed in U.S. Provisional Application Ser. No. 62/975,693, entitled “Nested and Parallel Feedback Control Loops For Ultra-Fine Measurements of Magnetic Fields From the Brain Using a Wearable MEG System” (Attorney Docket No. KERN-079).
The management control loop 54 manages the coarse feedback control loop 50 and fine feedback control loop 52 based on whether the fine magnetometers 26b are in-range or out-of-range, e.g., by considering coarse error signals SCERR from the coarse magnetometers 26a and ignoring fine error signals SFERR if the fine magnetometers 26b are out-of-range, and ignoring coarse error signals SCERR from the coarse magnetometers 26a and considering fine error signals SCERR from the fine magnetometers 26b if the fine magnetometers 26 are in-range. The management control loop 54 may monitor the spatial component of the total residual magnetic field BTOT and the overall behavior and history of the signal at each fine magnetometer 26b to determine whether or not the fine magnetometer 26b is in-range or out-of-range. It is noted that the spatial components of the total residual magnetic field BTOT at the fine magnetometers 26b may be substantially different from each other, and thus, some of the fine magnetometers 26b may be in-range, while other fine magnetometers 26b may be out-of-range.
With knowledge of whether each of the fine magnetometers 26b are in-range or out-of-range, the management control loop 54 may generally activate the fine feedback control loop 52 after initiating activation of the coarse feedback control loop 50. In this manner, as discussed above, the coarse feedback control loop 50 may coarsely control the actuated magnetic field BACT in a manner that at least partially cancels the outside magnetic field BOUT, and thus suppresses the total residual magnetic field BTOT at the fine magnetometers 26b to a baseline level, such that the at least one of magnetometers 26b comes in-range. The management control loop 54 may then activate the feedback control loop 52 to finely control the actuated magnetic field BACT in a manner that further suppresses the total residual magnetic field BTOT at the fine magnetometer(s) 26b that just came in-range to a lower level.
In one embodiment, the management control loop 54 strictly activates only the coarse feedback control loop 50 (e.g., if one of the fine magnetometers 26b is out-of-range) or only the fine feedback control loop (e.g., if all of the fine magnetometers 26 are in-range), but not both the coarse feedback control loop 50 and the fine feedback control loop 52 at the same time. In this case, the management control loop 54 will only consider coarse error signals SCERR from the coarse magnetometers 26a when the coarse feedback control loop 50 is active, and will only consider fine error signals SFERR from the fine magnetometers 26b when the fine feedback control loop 52 is active.
In another particularly preferred embodiment, however, the management control loop 54, at any given time, may not strictly activate only the coarse feedback control loop 50 or strictly activate only the fine feedback control loop 52, and thus, both of the coarse feedback control loop 50 and fine feedback control loop 52 may be at least partially activated. The management control loop 54 may choose to consider only the fine error signals SFERR from the fine magnetometers 26b that are in-range. In this case, the management control loop 54 may determine whether or not the fine magnetometer 26b is in-range, and performs a “sensor hand-off” procedure, and in particular, switches back and forth between consideration of a coarse error signal SCERR from any given coarse magnetometer 26a and consideration of a fine error signal SFERR from any given fine magnetometer 26b. It is understood that only some of the fine magnetometers 26b may be out-of-range at any given moment, so the sensor hand-off procedure can be from one, some, or all coarse magnetometers 26a to one, some, or all of the fine magnetometers 26b.
For example, if the management control loop 54 is currently considering a coarse error signal SCERR from a coarse magnetometer 26, and a previously unavailable fine magnetometer 26b is deemed to be in-range, the processor 30 may then ignore a coarse error signal SCERR from at least one coarse magnetometer 26a that is in proximity to the previously unavailable fine magnetometer 26b, and instead consider the more accurate fine error signal SFERR from this previously unavailable fine magnetometer 26b (in essence, passing or handing off detection of the total residual magnetic field BTOT from the coarse magnetometer(s) 26b to the fine magnetometer 26b).
On the contrary, if the management control loop 54 is currently considering a fine error signal SFERR from a fine magnetometer 26b, and the fine magnetometer 26b is subsequently deemed to fall out-of-range for any one of a variety of reasons (e.g., if the user 12, and thus the fine magnetometer 26b, gets too close to a power outlet, a fridge magnet, a cell phone, or perhaps if the user 12 turns their head so suddenly that the total residual magnetic field BTOT to which the fine magnetometer 26b varies too quickly), the management control loop 54 may then ignore the fine error signal SFERR from that fine magnetometer 26b, and instead consider the coarse error signal SCERR from at least one coarse magnetometer 26a in proximity to the now unavailable fine magnetometer 26b (in essence, passing or handing off detection of the total residual magnetic field BTOT from the fine magnetometer 26b to the coarse magnetometer 26a).
Thus, in this manner, the management control loop 54 may operate the fine feedback control loop 52 to control the actuated magnetic field BACT based on the fine error signals SFERR respectively output by fine magnetometers 26b as they come in-range. The management control loop 54 may operate the fine feedback control loop 52 to prevent control of the actuated magnetic field BACT based on the fine error signals SFERR respectively output by fine magnetometers 26b as they go out-of-range.
In an optional embodiment, the management control loop 54 may weight the fine magnetometers 26b, in which case, the management control loop 54 may not perform a “sensor hand-off” procedure, per se, but may assign a weight a to any given fine magnetometer 26b between a value 0 (no weight) and 1 (full weight). For example, the management control loop 54 may monitor different operating parameters of a fine magnetometer 26b to determine whether the fine magnetometer 26b is in a linear operating range, or outside of the linear operating range, but not saturated (non-linear operating range), or is saturated. If the fine magnetometer 26b is found to be in the linear operating range, the weighting a assigned to the fine magnetometer 26b can be 1 (i.e., full weight); if the fine magnetometer 26b is found to be saturated, the weighting a assigned to the fine magnetometer 26b can be 0 (i.e., no weight); and if the fine magnetometer 26b is found to be in the non-linear operating range, the weighting a assigned to the fine magnetometer 26b can be between 0 and 1 (i.e., partial weight), depending on how close the fine magnetometer 26b is to saturation.
As discussed above, the management control loop 54 is configured for correcting factors that can change more slowly over time to optimize the cancellation of the outside magnetic field BOUT. For example, the management control loop 54 may be configured for implementing adaptions to slow changes of the coarse feedback control loop 50 and fine feedback control loop 52 over time. The management control loop 54 is configured for identifying and determining parameters and coefficients of the signal acquisition unit 18 and the outside magnetic field BOUT. The management control loop 54 is configured for employing computational algorithms to determine unknown parameters from the coarse error signals SCERR and fine error signals SFERR output by the coarse magnetometers 26a and fine magnetometers 26b, such as fitting of physical and calibrated mathematical and numerical models to the coarse error signals SCERR and fine error signals SFERR to identify missing or insufficiently known coefficients and parameters. Such parameters and coefficients can include offset and gain coefficients for the coarse magnetometers 26a, gain constants for the fine magnetometers 26b, actuator gains and offsets for the set of magnetic field actuators 28, electronics time delay latency coefficients in the coarse feedback control loop 50 and fine feedback control loop 52 (i.e., the amount of time between generating the coarse error signal SCERR or fine error signal SFERR and activating the set of magnetic field actuators 28), and other parameters of the signal acquisition unit 18. The management control loop 54 may determine coefficients and parameters for different temporal and spatial ranges. Likewise, the gain that the set of magnetic field actuators 28 may have on the coarse magnetometers 26a and fine magnetometers 26b may differ with the placement and location offset of magnetic field actuators 28 (e.g., as the head of the user 12 moves or the support structure 24 deforms). The management control loop 54 may identify at least one, some, or all of the coefficients or parameters over these changing conditions.
In one exemplary instance, a mathematical and numerical model of the signal acquisition unit 18, or a portion thereof, has some coefficients or parameters that are considered poorly or insufficiently known. In another exemplary instance, a mathematical and numerical model of the signal acquisition unit 18 does not have a predetermined structure, and the coefficients or parameters consist of transfer functions or linear mappings from one set of signals to another. The management control loop 54 may compare the response of a structured or unstructured model of the signal acquisition unit 18 to the measurements from the coarse magnetometers 26a and fine magnetometers 26b, and the coefficients or parameters may be varied until any disagreement between the mathematical model of the signal acquisition unit 18 and the actual measured signals is decreased. The coefficients or parameters of the mathematical model that achieve such a decrease in disagreement are the estimated parameters of the signal acquisition unit 18 (meaning, if the mathematical model with selected parameter values x, y, and z best matches the actual measured behavior of the system, then the values x, y, and z are a system identification estimate of the poorly or insufficiently known coefficients or parameters of the system). In determining the coefficients or parameters of the signal acquisition unit 18, the management control loop 54 may employ weighted least squares, observer filters, Kalman filters, Wiener filters, or other filters. The management control loop 54 may employ time domain, frequency domain, recursive techniques, parametric and non-parametric methods, linear and nonlinear optimization techniques including gradient descent, matrix methods, convex methods, non-convex methods, neural networks, genetic algorithms, fuzzy logic, and machine learning methods.
The management control loop 54 may perform calibration techniques prior to operating the neural activity measurement system 10, or calibration techniques may be performed in real-time as the neural activity measurement system 10 operates. For example, prior to usage, the signal acquisition unit 18 may be calibrated by applying a known magnetic field in a controlled shielded setting (e.g., to characterize the coarse magnetometers 26a for their offsets and gain measurements). However, the properties of coarse magnetometers 26a, fine magnetometers 26b, or set of magnetic field actuators 28 may vary due to environmental variations, such as, e.g., variations in temperature, laser power (for magnetometers that utilize lasers), motion or deformation of the support structure 24, or other deformations, such as bending of the coarse magnetometers 26a, fine magnetometers 26b, or offset of magnetic field actuators 28 due to temperature or mechanical stresses. Thus, in addition to performing calibrations ahead of time, the management control loop 54 may perform calibrations techniques during system operation. For example, if the offsets and gains of the coarse magnetometers 26a change during usage of the neural activity measurement system 10, the management control loop 54 may estimate the offsets and gains of the coarse magnetometers 26a in real time (i.e., as the neural activity measurement system 10 is running), e.g., by estimating and comparing the offset of one coarse magnetometer against the measurements of other coarse or fine magnetometers. Further details discussing the calibration of coarse magnetometers are disclosed in U.S. Provisional Application Ser. No. 62/975,709, entitled “Self-Calibration of Flux Gate Offset and Gain Drift To Improve Measurement Accuracy Of Magnetic Fields From the Brain Using a Wearable MEG System” (Attorney Docket No. KERN-078), which is expressly incorporated herein by reference.
It should be appreciated that, in the case where the signal acquisition unit 18 comprises multiple sets of magnetic field actuators 28 and processors 30, the components, along with the coarse feedback control loop 50, fine feedback control loop 52, and management control loop 54, illustrated in
Although the total residual magnetic field BTOT may be suppressed to a level that allows the ultra-fine measurements of the MEG magnetic field BMEG emanating from the brain 14 of the user 12 to be taken by the fine magnetometers 26b, some portion of the outside magnetic field BOUT will likely remain in the total residual magnetic field BTOT measured by the fine magnetometers 26b, and thus, will be considered environmental magnetic noise to the relatively weak MEG signals SMEG contained in the measured total residual magnetic field BTOT.
Significantly, although the measurement errors of fine magnetometers 26b are relatively small, the processor 30 is configured for distinguishing the portion of the measured total residual magnetic field BTOT-MEAS that corresponds to the true MEG magnetic field BMEG-TRUE (i.e., the true magnetic field that emanates from the head of the user 12 due to neural activity in the brain 14) and the portion of the measured total residual magnetic field BTOT-MEAS that does not correspond to the true MEG magnetic field BMEG-TRUE by employing a combination of three signal discriminating techniques, thereby maximizing the accuracy of the measurements of these fine magnetometers 26b.
In particular, referring to
For example, as illustrated in
With regard to the size (vertical) axis of
Thus, it can be appreciated from
In one embodiment, the processor 30 accomplishes this by transforming the measured total residual magnetic field BOUT-MEAS from a time domain into the frequency domain (e.g., using a Fast Fourier Transform (FFT), and eliminating the content of the measured total residual magnetic field BTOT-MEAS corresponding to the frequency components having peak amplitudes that are above and below the pico tesla (pT) range. For example, as illustrated in
The processor 30 may accomplish this by filtering the measured total residual magnetic field BTOT-MEAS at these frequency components, and cancelling the outside magnetic field BOUT at these frequency components (e.g., by generating cancellation control signals based on this determination that are output to the set of magnetic field actuators 28 and actuating the set of magnetic field actuators 28 in accordance with these cancellation control signals. Alternatively, the processor 30 may eliminate the content of the measured total residual magnetic field BTOT-MEAS corresponding to certain frequency components by filtering these frequency components out of the measured total residual magnetic field BTOT-MEAS during a post-cancellation step.
It should be appreciated that the strength thresholds of the outside magnetic field BOUT, MEG magnetic field BMEG, and measurement noise δ may vary. For example, the distance from the brain 14 and the location of the fine magnetometers 26b may change, because different individuals may have different skull, skin (i.e., scalp), and hair thicknesses, and because the fine magnetometers 26b may be in direct contact with the scalp or may be set back from the scalp (e.g., to accommodate additional elements in the wearable signal processing unit 18, or for thermal management reasons). As the distance of a fine magnetometer 26b and the brain 14 increases, the strength of the MEG magnetic field BMEG at that fine magnetometer 26b decreases, and thus, the strength threshold at which moderate strength components of the measured total residual magnetic field BOUT-MEAS is selected relative to the too-strong or too-weak strength components of the measured total residual magnetic field BOUT-MEAS may be modified to account for the collective distance between the fine magnetometers 26b and the brain 14 of the user 12.
With regard to the temporal frequency (horizontal) axis of
The outside magnetic field BOUT also comprises temporal frequency components at 60 Hz due to time-varying electromagnetic radiation emanating from electrical outlets and sockets, electrical wires or connections in the wall, and everyday electrical equipment when the user 12 is in a home, work, or laboratory setting in the United States, as well as at multiples of 60 Hz due to non-linear interactions between the electromagnetic radiation and environment (as the electromagnetic field couples with everyday objects, e.g., metal spars in a chair or table or refrigerator) that lead to frequency doublings, triplings, etc.
It should be appreciated that the harmonic components in the outside magnetic field BOUT may be in multiples of a frequency that differs from 60 Hz. For example, the harmonic components may be in multiples of 50 Hz if the home, work, or laboratory setting is in Europe. Also, due to variations and imperfections in the power supply to electronics, the harmonic components in the outside magnetic field BOUT may be in multiples of a frequency that is exactly 60 Hz (or 50 Hz). Furthermore, due to coupling of the electromagnetic radiation with the environment, some frequency spread may occur, such that the finite frequency bands centered at (or approximately at) 60 Hz, 120 Hz, 180 Hz, etc. (or 50 Hz, 100 Hz, 150 Hz, etc.) occur in the outside magnetic field BOUT.
Based on this knowledge, the processor 30 may reduce the content of the outside magnetic field BOUT and measurement noise δ in the measured total residual magnetic field BTOT-MEAS by eliminating the content of the measured total residual magnetic field BTOT-MEAS corresponding to temporal frequency components in the range of DC to a few Hertz, in the range of thousands of Hertz, and also at harmonic temporal frequencies of 60 Hz (or 50 Hz).
The processor 30 may accomplish this by filtering the measured total residual magnetic field BTOT-MEAS at these temporal frequency components, and cancelling the outside magnetic field BOUT at these temporal frequency components (e.g., by generating cancellation control signals based on this determination that are output to the set of magnetic field actuators 28 and actuating the set of magnetic field actuators 28 in accordance with these cancellation control signals. Further details discussing cancelling the outside magnetic field BOUT at selected temporal frequency components are disclosed in U.S. Provisional Application Ser. No. 62/975,693, entitled “Nested and Parallel Feedback Control Loops For Ultra-Fine Measurements of Magnetic Fields From the Brain Using a Wearable MEG System” (Attorney Docket No. KERN-079), which is expressly incorporated herein by reference. Alternatively, the processor 30 may eliminate the content of the measured total residual magnetic field BTOT-MEAS corresponding to certain temporal frequency components by filtering these temporal frequency components out of the measured total residual magnetic field BTOT-MEAS during a post-cancellation step.
With regard to the spatial frequency (diagonal) axis of
Just like a magnetic field can have temporal frequency components (e.g., slow or less than 5 Hz), three-dimensional magnetic fields also have spatial frequency components (long spatial wavelength (e.g. greater than 1 meter) corresponds to low spatial frequency (e.g., less than 1 cycle/meter)). Thus, a magnetic field may oscillate over short distances in space (short wavelength) or oscillate over long distances (long wavelengths).
The Earth's magnetic field has a low spatial frequency. For example, magnetic North is basically the same on one side of a room as it is on the other side of the room. When the Earth's magnetic field interacts with everyday objects that have magnetizable components; for example, a chair leg, table spar or a beam in the wall of a room that is composed of metal, such as ferrous iron, that is magnetically responsive, such everyday objects may modify the Earth's magnetic field (a bending and curvature of Earth's magnetic field) or equivalently the magnetic flux lines. Thus, in a home, office, or laboratory environment, the Earth's magnetic field may spatially vary in the vicinity of the magnetizable metals or other magnetizable materials. However, the Earth's magnetic field is only modified modestly by magnetizable components, and unless the signal acquisition unit 18 is very close to a magnetizable component (e.g., if the user 12 places their head right next to an iron leg), the modestly modified Earth's magnetic field will be almost constant across the head of the user 12 or, if more accuracy is desired, may be represented accurately by a constant (0th order) component plus a linear (1st order) component. If a compass was held at one side or the other side of the head of the user 12, the direction and strength of the Earth's magnetic field, would be about the same for both sides of the user of the user 12. Hence, the resulting spatial frequencies in an outside magnetic field BOUT corresponding to the Earth's magnetic field, even in an indoor or outdoor setting where there are magnetizable materials in the vicinity, are still typically small.
Likewise, even though the time-varying electromagnetic radiation emanating from electrical outlets and sockets, electrical wires or connections in the wall, and everyday electrical equipment when the user 12 is in a home, work, or laboratory setting has fast varying harmonic temporal frequency components, the spatial short wavelength components of this electromagnetic radiation quickly dissipates with distance from the source or sources of the electromagnetic radiation. Thus, as long as the signal acquisition unit 18 is not adjacent to the source of the electromagnetic energy, then the spatial short wavelength components of the electromagnetic radiation will have dissipated by the time the electromagnetic radiation has reach the head of the user 12. Thus, the spatial frequency components of the outside magnetic field BOUT will generally be relatively low.
In contrast, the spatial frequency components of the MEG magnetic field BMEG will generally be higher at the head of the user 12 than those of the outside magnetic field BOUT. Neurons in the brain 16 of the user 12 that produce the electrical currents are packed closely together, such that they create a MEG magnetic field BMEG with short wavelength components.
Based on this knowledge, the processor 30 may reduce the content of the outside magnetic field BOUT in the measured total residual magnetic field BTOT-MEAS by eliminating the content of the measured total residual magnetic field BTOT-MEAS corresponding to low spatial frequency components. For example, referring to
The processor 30 collectively processes the spatial components of the total residual magnetic field BTOT-MEAS measured by the magnetometers 26 in a manner that cancels the content of the outside magnetic field BOUT from the measured total residual magnetic field BTOT-MEAS. For example, the spatial components of the magnetometers 26 may be averaged to acquire a DC level that can then be individually subtracted from the spatial components of the measured total residual magnetic field BTOT-MEAS, thereby reducing the content of the outside magnetic field BOUT in the measured total residual magnetic field BTOT-MEAS.
Referring back to
It should be appreciated that although the processor 30 may be configured for distinguishing the MEG magnetic field BMEG, outside magnetic field BOUT, and measurement noise δ based on any combination of strength, temporal frequency, and spatial frequency, and in any order, the processor 30 may be configured for selecting the combination and order of strength, temporal frequency, and spatial frequency on which to distinguish the MEG magnetic field BMEG, outside magnetic field BOUT, and measurement noise δ based on certain criteria.
There may be conditions where the strength components, temporal frequency components, or spatial frequency components of the MEG magnetic field BMEG coincide with the strength components, temporal frequency components, or spatial frequency components of the outside magnetic field BOUT or measurement noise δ, in which case, the processor 30 may not opt to not eliminate content of the outside magnetic field BOUT and/or measurement noise δ based on the coinciding strength, temporal frequency, and/or spatial frequency.
As one example, in the case where MEG magnetic field BMEG and the outside magnetic field BOUT are distinguished based on the strength or temporal frequency, eliminating content of the measured total residual magnetic field BTOT-MEAS may inadvertently eliminate underlying content of the MEG magnetic field BMEG corresponding to frequency components that coincide with the frequency components at which the content of the measured total residual magnetic field BTOT-MEAS has been eliminated, e.g., at 60 Hz or 120 Hz.
In this case, the content of the measured total residual magnetic field BTOT-MEAS corresponding to these frequency components may instead be retained in the measured total residual magnetic field BTOT-MEAS, and may be eliminated from the measured total residual magnetic field BTOT-MEAS in a different regime. For example, it is likely that the content of the outside magnetic field BOUT corresponding to the same frequency components of the underlying content of the MEG magnetic field BMEG has been contributed by electromagnetic radiation from electrical equipment or power sources that has a low spatial frequency in contrast to the high spatial frequency of the MEG magnetic field BMEG.
Thus, the processor 30 may opt to distinguish the MEG magnetic field BMEG and the outside magnetic field BOUT based on spatial frequency, in which case, it can eliminate at least portion of the content of the outside magnetic field BOUT from the measured total residual magnetic field BTOT-MEAS without eliminating the content of the MEG magnetic field BMEG as discussed above.
As another example, due to the interaction between the electromagnetic radiation and the environment, the strength of the harmonic frequency components in the outside magnetic field BOUT may have not always be at a relatively high amplitude, but can be at a relatively low amplitude commensurate with the strength of the MEG magnetic field BMEG. In such case, the MEG magnetic field BMEG and the outside magnetic field BOUT may not be distinguished from each other based on strength, as discussed above. The MEG magnetic field BMEG and the outside magnetic field BOUT may also not be distinguished based on temporal frequency, since the harmonic frequency components of the outside magnetic field BOUT are likely to coincide with frequency components of the MEG magnetic field BMEG.
However, the MEG magnetic field BMEG and the outside magnetic field BOUT may be distinguished from each other based on spatial frequency even when the strength and harmonic frequency components of the outside magnetic field BOUT are commensurate with the strength and frequency components of the MEG magnetic field BMEG. Thus, the processor 30 may opt to distinguish the MEG magnetic field BMEG and the outside magnetic field BOUT based on spatial frequency, in which case, it can eliminate at least portion of the content of the outside magnetic field BOUT from the measured total residual magnetic field BTOT-MEAS without eliminating the content of the MEG magnetic field BMEG as discussed above.
The processor 30 may be configured for dynamically selecting the combination and order of strength, temporal frequency, and spatial frequency on which to distinguish the MEG magnetic field BMEG, outside magnetic field BOUT, and measurement noise δ based on criteria other than the inadvertent coincidence of strength, temporal frequency, or spatial frequency between the MEG magnetic field BMEG and the outside magnetic field BOUT or measurement noise δ.
For example, distinguishing between the MEG magnetic field BMEG and the outside magnetic field BOUT and measurement noise δ based on temporal frequency relies on priori knowledge that the MEG magnetic field BMEG and the outside magnetic field BOUT and measurement noise δ have certain dominant temporal frequency components. In contrast, while the preferred embodiment of distinguishing between the MEG magnetic field BMEG and the outside magnetic field BOUT and measurement noise δ based on strength relies on analyzing the frequency components of the measured total residual magnetic field BTOT-MEAS in the frequency domain, such technique does not rely on prior knowledge that the highest strength of the MEG magnetic field BMEG and the outside magnetic field BOUT and measurement noise δ is at certain temporal frequency components.
Although there may be some expectation that certain frequency components of the measured total residual magnetic field BTOT-MEAS analyzed in the frequency domain will be dominant, and thus may coincide with the same temporal frequency components that the measured total residual magnetic field BTOT-MEAS will be used to distinguish between the MEG magnetic field BMEG and the outside magnetic field BOUT and measurement noise δ based on temporal frequency, thereby may be unexpected dominant frequency components in the frequency domain of the measured total residual magnetic field BTOT-MEAS that do not coincide with the temporal frequency components that will be, or have been, used to distinguish between the MEG magnetic field BMEG and the outside magnetic field BOUT and measurement noise δ based on temporal frequency (e.g., if at least a portion of the content of the outside magnetic field BOUT corresponds electromagnetic radiation having temporal frequency components that are not at DC or the 60 Hz harmonic components).
Thus, the processor 30 may opt to first distinguish MEG magnetic field BMEG and the outside magnetic field BOUT and measurement noise δ based on temporal frequency, such that at least some of the content of the outside magnetic field BOUT and measurement noise δ is eliminated from the measured total residual magnetic field BTOT-MEAS. If the measured total residual magnetic field BTOT-MEAS, after such content has been eliminated, is still too high, the processor 30 may opt to then distinguish MEG magnetic field BMEG and the outside magnetic field BOUT and measurement noise δ based on strength, such that more content of the outside magnetic field BOUT and measurement noise δ is eliminated from the measured total residual magnetic field BTOT-MEAS. If the measured total residual magnetic field BTOT-MEAS, after such additional content has been eliminated, is still too high, the processor 30 may opt to then distinguish MEG magnetic field BMEG and the outside magnetic field BOUT and measurement noise δ based on spatial frequency, such that even more content of the outside magnetic field BOUT and measurement noise δ is eliminated from the measured total residual magnetic field BTOT-MEAS.
The processor 30 may opt to dynamically select whether or not to eliminate content of the outside magnetic field BOUT or measurement noise δ from the measured total residual magnetic field BTOT-MEAS based on practical considerations, even after properly distinguishing the MEG magnetic field BMEG, outside magnetic field BOUT, and measurement noise δ. For example, due to complex factors, the outside magnetic field BOUT may comprise strong frequency components at 60 Hz, 120 Hz, and 240 Hz, but a weak frequency component at 180 Hz. In this instance, the processor 30 may opt to eliminate the content of the outside magnetic field BOUT from the measured total residual magnetic field BTOT-MEAS at 60 Hz, 120 Hz, and 240 Hz, but not at 240 Hz if it is deemed that attempting to eliminate the content of the outside magnetic field BOUT from the measured total residual magnetic field BTOT-MEAS at 240 Hz would add more noise to the measured total residual magnetic field BTOT-MEAS than the noise created by the 240 Hz frequency component of the outside magnetic field BOUT.
Referring back to
Thus, the processor 30 is configured for correcting the measurement errors in the environmental magnetic field BENV component of the total residual magnetic field measurements BTOT-MEAS, thereby increasing the accuracies of the estimates of the total residual magnetic field BTOT at the fine magnetometers 26b. As a result of reducing measurements errors associated with the outside magnetic field BOUT component in the total residual magnetic field measurements BTOT-MEAS, the outside magnetic field BOUT may be more accurately cancelled, thereby more effectively suppressing the total residual magnetic field BTOT at the fine magnetometers 26b to bring the fine magnetometers 26b in-range. Furthermore, as a result reducing measurements errors associated with the MEG magnetic field BMEG component in the total residual magnetic field measurements BTOT-MEAS, the MEG magnetic field BMEG may be more accurately determined. In effect, the physical (true) portion of the MEG magnetic field BMEG component of the measured total residual magnetic field BTOT-MEAS and the non-physical (error) portion of the MEG magnetic field BMEG component of the measured total residual magnetic field BTOT-MEAS is distinguished, as represented by the union space 68 between the oval 60 and the bottom triangle 64.
To this end, the processor 30 is configured for inferring total residual magnetic field estimates BTOT-EST at the magnetometers 26 by (1) acquiring the measurements of the total residual magnetic field BTOT-MEAS from the magnetometers 26 (i.e., the coarse error signals SCERR and/or fine error signals SFERR); (2) determining the known actuated magnetic field BACT-KNOWN at the magnetometers 26 based on a known profile of the set of magnetic field actuators 28 and the actuation strengths of the magnetic field actuators 28; (3) generating a generic model of the environmental magnetic field BENV-MOD in the vicinity of the magnetometers 26; (4) constraining the environmental magnetic field model BENV-MOD to generate a Maxwell-constrained model of the environmental magnetic field BENV-MAXWELL that satisfies Maxwell's equations; (5) parameterizing the Maxwell-constrained environmental magnetic field model BENV-MAXWELL based on the measured total residual magnetic field BTOT-MEAS measured by the magnetometers 26 and the known actuated magnetic field BACT-KNOWN at the magnetometers 26 to generate a parameterized environmental magnetic field model BENV-PAR (representative of the true environmental magnetic field model BENV in the vicinity of the magnetometers 26); (6) determining the environmental magnetic field estimates BENV-EST at the magnetometers 26 based on the parameterized environmental magnetic field model BENV-PAR; and (7) determining the total residual magnetic field estimates BTOT-EST at the magnetometers 26 based on the known actuated magnetic field BACT-KNOWN at the magnetometers 26 and the environmental magnetic field estimates BENV-EST at the magnetometers 26.
As described below, the total residual magnetic field estimates BTOT-EST are inferred at both the coarse magnetometers 26a and fine magnetometers 26b based on total residual magnetic field measurements BTOT-MEAS acquired from both the coarse magnetometers 26a and fine magnetometers 26b, but in alternative embodiments, the total residual magnetic field estimates BTOT-EST may be inferred at only the coarse magnetometers 26a based on total residual magnetic field measurements BTOT-MEAS acquired from both the coarse magnetometers 26a and fine magnetometers 26b, or may be inferred at only the fine magnetometers 26b based on total residual magnetic field measurements BTOT-MEAS acquired from only the fine magnetometers 26b.
With regard to acquiring the total residual magnetic field measurements BTOT-MEAS from the magnetometers 26, in an exemplary embodiment, an Nc number of coarse magnetometers 26a respectively at an Nc number of locations may collect an Nc×K coarse measurements of the total residual magnetic field BTOT-MEASC over time in accordance with the discretized matrix:
Similarly, an NF number of fine magnetometers 26b respectively at an NF number of locations collect an NF×K fine measurements of the total residual magnetic field KTOT-MEASF over time in accordance with the discretized matrix:
One of ordinary skill in the art of control and signal processing will recognize that the timing of the coarse Nc×K total residual magnetic field measurements BTOT-MEASC taken by the coarse magnetometers 26a and the fine NF×K total residual magnetic field measurements BTOT-MEASF taken by the fine magnetometers 26b need not be the same, and that the coarse Nc×K total residual magnetic field measurements BTOT-MEASC, NF×K total residual magnetic field measurements BTOT-MEASF, and M×K actuations of the actuated magnetic field BACT may be performed at the same time and may be non-synchronized.
Each of the coarse Nc×K total residual magnetic field measurements BTOT-MEASC, NF×K total residual magnetic field measurements BTOT-MEASF, and M×K actuations of the actuated magnetic field BACT is known imperfectly. For example, although each of the fine NF×K total residual magnetic field measurements BTOT-MEASF, may have a relatively high accuracy, each of the fine magnetometers 26b still have a measurement variance on the order of picoteslas (pT). In contrast, each of the coarse Nc×K total residual magnetic field measurements BTOT-MEASC has a relatively low accuracy, and in particular, each of the coarse magnetometers 26a may have a much higher measurement variance on the order of microteslas (μT) or tens or hundreds of microteslas (μT).
For the purposes of the following discussion, the coarse Nc×K total residual magnetic field measurements BTOT-MEASC and fine NF×K total residual magnetic field measurements BTOT-MEASF can be consolidated into an N× K number of total residual magnetic field measurements BTOT-MEAS (the sum of the Nc number of coarse magnetometers 26a and the NF number of fine magnetometers 26b (if available)), such that equations [1a] and [1b] reduces to:
Assuming that the magnetometers 26 are vector magnetometers for respectively measuring the x-, y-, and z-components of the total residual magnetic field measurements BTOT-MEAS, equation [1] can be expressed as a vector {right arrow over (BTOT-MEAS)}(x, y, z, t) that varies over space and time, where x, y, z are the three cardinal directions, and t is time that varies over space and time.
The total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) at the locations of an N number of magnetometers 26 may be given as:
As briefly discussed above, the processor 30 may determine the known actuated magnetic field BACT-KNOWN at the magnetometers 26 based on a known profile of the set of magnetic field actuators 28 and the actuation strengths of the magnetic field actuators 28. In an exemplary embodiment, an M number of the magnetic field actuators 28 may apply an M×K actuations of the actuated magnetic field BACT over time in accordance with the discretized matrix:
Assuming that the set of magnetic field actuators 28 comprises a triad of uniform magnetic field actuators 28a-28c (M=3) for respectively generating x-, y-, and z-components of the actuated magnetic field BACT to cancel the outside magnetic field BOUT in all three dimensions, the actuated magnetic field BACT can be defined as a vector {right arrow over (BACT)}(x, y, z, t) that varies over space and time.
The set of magnetic field actuators 28 respectively have an M number of actuation strengths in the form of a vector (t) (one for each magnetic field actuator 28) and a matrix of influence R by the actuation strength vector (t) to the actuated magnetic field {right arrow over (BACT)}(x, y, z, t) at the N number of magnetometers 26, as follows:
The matrix of influence R may be generated using mathematical or numerical modeling (e.g., by simulating the magnetic field emanating from each of the magnetic field actuators 28 to different spatial locations, e.g., at the magnetometers 26) or by the performance of calibration measurements ahead of time (i.e., generate a nominal actuated magnetic field and measure the actuated magnetic field at different spatial locations, e.g., at the magnetometers 26) that quantifies the profile of the actuated magnetic field BACT generated by each of magnetic field actuators 28, and therefore defines the influence of each magnetic field actuator 28 at the location of each magnetometer 26. The resulting actuated magnetic field at the locations of the magnetometers 26 will linearly scale with the actuation strength vectors j(t) of the magnetic field actuators 28, such that a known actuated magnetic field {right arrow over (BACT-KNOWN)}(x, y, z, t) that varies over space and time at the N number of magnetometers 26 may be given as:
In this particular embodiment, the minute contribution of the MEG magnetic field BMEG is ignored for now for purposes of simplicity, such that the environmental magnetic field BENV, generic environmental magnetic field model BENV-MOD Maxwell-constrained environmental magnetic field model BENV-MAXWELL, and parameterized environmental magnetic field model BENV-MAXWELL can be respectively replaced with the outside magnetic field BOUT, a generic outside magnetic field model BOUT-MOD, a Maxwell-constrained outside magnetic field model BOUT-MAXWELL, and a parameterized outside magnetic field model BOUT-MAXWELL. In this case, the physical portion of the MEG magnetic field BMEG component in the measured total residual magnetic field BTOT-MEAS and the non-physical (error) portion of the MEG magnetic field BMEG component in the measured total residual magnetic field BTOT-MEAS are not distinguished. Rather, only the measurement errors associated with the outside magnetic field BOUT component in the total residual magnetic field measurements BTOT-MEAS will be reduced, such that the outside magnetic field BOUT may be more accurately cancelled, thereby more effectively suppressing the total residual magnetic field BTOT at the fine magnetometers 26b to bring the fine magnetometers 26b in-range.
As briefly discussed above, the processor 30 may generate the environmental magnetic field model BENV-MOD, and in this particular case the outside magnetic field model BOUT-MOD, in the vicinity of the magnetometers 26. In particular, on the length scale of the signal acquisition unit 18, the outside magnetic field BOUT may assume to have certain physical properties. The processor 30 may generate the generic outside magnetic field model BOUT-MOD in the vicinity of the magnetometers 26 based on these assumed physical properties in any one of a variety of manners, but in the illustrated embodiment, the processor 30 models the outside magnetic field BOUT as a function of space by employing one or more basis functions. In one embodiment, the processor 30 models the outside magnetic field BOUT by employing basis functions having a linear spatial dependence. For example, one basis function may have a uniform (0th order) components and linear (first order) spatial components (i.e., the slope). Second order non-linear spatial components can be ignored, although in alternative embodiments, basis functions with non-linear spatial dependence, or other types of modeling that one of ordinary skill in the art of signal processing, system identification, or control will recognize will serve the same purpose (such as other types of modes or bases, including singular values, eigenvectors, or bases collected from data such as collected by proper orthogonal decomposition or by other fitting methods).
Assuming that the outside magnetic field BOUT can be modeled with only 0th order and 1st order components, a time-varying and spatially-varying generic model of the magnetic field model {right arrow over (BOUT-MOD)}(x, y, z, t) is:
where O(∥x,y,z∥2) means that the neglected higher order terms produce an error that scales as ∥x,y,z∥2, which is the size of the vector to the second power. As described above, this error is practically small for the outside magnetic field BOUT. Hence, the x-directional component BxOUT-MOD(x, y, z, t) of the magnetic field model {right arrow over (BOUT-MOD)}(x, y, z, t) has a 0th order component that is characterized by the time-varying basis function αx(t) and 1st order spatial components that linearly vary in the space (x, y, and z) and are respectively characterized by time varying basis functions αxx(t)x, αxy(t)y, and αxz(t)z; the y-directional component ByOUT-MOD(x, y, z, t) of the magnetic field model {right arrow over (BOUT-MOD)}(x, y, z, t) has a 0th order component that is characterized by the time-varying basis function αy(t) and 1st order spatial components that linearly vary in the space (x, y, and z) and are respectively characterized by time varying basis functions αyx(t)x, αyy(t)y, and αyz(t)z; and the y-directional component BzOUT-MOD(x, y, z, t) of the magnetic field model {right arrow over (BOUT-MOD)}(x, y, z, t) has a 0th order component that is characterized by the time-varying basis function αz(t) and 1st order spatial components that linearly vary in the space (x, y, and z) and are respectively characterized by time varying basis functions αzx(t)x, αzy(t)y, and αzz(t)z.
Thus, a total of 12 initial basis functions (i.e., αx(t), αxx(t)x, αxy(t)y, αxz(t)z, αy(t), αyx(t)x, αyy(t)y, αyz(t)z, αz(t), αzx(t)x, αzy(t)y, αzz(t)z) characterizes the magnetic field model {right arrow over (BOUT-MOD)}(x, y, z, t). As will be described in further detail below, a coefficient vector (t)=[γ1(t), γ2 (t), . . . γ12 (t)] respectively associated with these basis functions can be estimated based on the total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) acquired from the magnetometers 26. Higher order spatial components, such as second order terms in space like x2, xy, and z2, and third, fourth, and fifth order terms, etc., for this exemplary instance are assumed negligible.
As briefly discussed above, the processor 30 may constrain the environmental magnetic field model BMEG-MOD to generate a Maxwell-constrained environmental magnetic field model BMEG-MAXWELL that satisfies Maxwell's equations, and in this particular case, may constrain the outside magnetic field model {right arrow over (BOUT-MOD)}(x, y, z, t) to generate a Maxwell-constrained environmental magnetic field model {right arrow over (BOUT-MAXWELL)}(x, y, z, t) that satisfies Maxwell's equations. In particular, using the known physicals of Maxwell's equations, the processor 30 reduce the number of coefficients to be estimated. As a result, a smaller number of coefficients are estimated with the same number of available measurements, and the resulting accuracy of the estimation can improve for two reasons: firstly, because from the total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) that are available less coefficients need to be estimated; and secondly because exploiting Maxwell's equations as disclosed can allow measurement errors that reflect a situation that is not physically possible (does not satisfy Maxwell's equations) to be eliminated, and thus only the errors (e.g., errors in the generic outside magnetic field model {right arrow over (BTOT-MOD)}(x, y, z, t)) that satisfy Maxwell's equations remain. Specifically, the outside magnetic field {right arrow over (BOUT)}(x, y, z, t) (or any magnetic field) can be expressed as:
{right arrow over (BOUT)}(x,y,z,t)={right arrow over (BPHYSICAL)}(x,y,z,t)+{right arrow over (BNON-PHYSICAL)}((x,y,z,t), [7]
where
{right arrow over (BPHYSICAL)}(x, y, z, t) satisfies Maxwell's equations and can occur, and
{right arrow over (BNON-PHYSICAL)}(x, y, z, t) does not satisfy Maxwell's equations and cannot occur.
Measurement errors can occur in all directions, and can have modes both along {right arrow over (BPHYSICAL)}(x, y, z, t) and {right arrow over (BNON-PHYSICAL)}(x, y, z, t). Using Maxwell's equations, the processor 30 may distinguish the modes of the outside magnetic field {right arrow over (BOUT)}(x, y, z, t) that are physically possible, and the modes of the outside magnetic field {right arrow over (BOUT)}(x, y, z, t) that are physically impossible. Thus, employing Maxwell's equations to the generic outside magnetic field model {right arrow over (BOUT-MOD)}(x, y, z, t) eliminate errors along the physically not possible direction.
Maxwell's equations include Gauss' Law, Gauss' Law for Magnetism, Faraday's Law, and Ampere's-Maxwell's Law.
Gauss' Law describes the relationship between a static electric field and the electric charges, and in particular, states that the net outflow of the electrical field through any closed surface is proportional to the charge enclosed by the surface, in accordance with:
where “∇·” is a divergence operator, E is the electric field, ρ is the total charge per unit volume, and ∈0 is the permittivity of free space.
Gauss's Law for Magnetism states that there are no magnetic charges (also called “magnetic monopoles”), but instead the magnetic field is generated by a dipole, such that the net outflow of the magnetic field through any closed surface is zero, in accordance with:
∇·B=0, [8b]
where “∇·” is a divergence operator and B is the magnetic field.
Faraday's Law describes the relationship between a time-varying magnetic field and an electric field, and states that, the work per unit charge required to move a charge around a closed loop equals the rate of change of the magnetic flux through the enclosed surface, in accordance with:
where “∇×” is the curl operator, E is the electric field, and
is the change in magnetic field per unit time.
Ampere's-Maxwell's Law states that magnetic fields can be generated by changing electric fields, and states that the magnetic field induced around any closed loop is proportional to the electric current and the displacement current (proportional to the rate of change of electric flux) through the enclosed surface, in accordance with:
where “∇×” is the curl operator, B is the magnetic field,
is the change in electric field per unit time, and J is the current density, μ0 is the permeability of free space, and ∈0 is the permittivity of free space.
The four Maxwell's equations can be used to constrain the 1st order coefficients αxx, αyy, and αzz to:
αxx+αyy+αzz=0; and [9a]
the remaining 1st order coefficients αxy, αyx, αxz, αzx, αyz, and αzy, (assuming electromagnetic terms are small for the measurement of the frequencies of interest) to:
−αxy+αyx=0; [9b]
αxz−αzx=0; and [9c]
−αyz+αzy=0. [9d]
In total, these are four equations (equations [9a]-[9d]) for 12 coefficients, which can be represented in matrix form as:
M
(t)=0, [10]
where =[αx,αy, . . . αzz].
After solving equations [9a]-[9d], there will be 8 degrees of freedom left. Indeed, the null space of the matrix M yields all possible coefficients at any time, by the equation:
(t)=Γ, [11]
where =[γ1, γ2, . . . γ8] contains only 8 free coefficients instead of 12 free coefficients, and F is a null matrix of the matrix M. Thus, in an exemplary embodiment, there are 8 free parameters that reduces the generic outside magnetic field model {right arrow over (BOUT-MOD)}(x, y, z, t) of equation [6] to a Maxwell-constrained outside magnetic field model {right arrow over (BOUT-MAXWELL)}(x, y, z, t) with eight basis functions corresponding to the Maxwell-constrained coefficient vector (t)=[γ1(t), γ2(t), . . . γ8(t)]. The Maxwell-constrained outside magnetic field model {right arrow over (BOUT-MAXWELL)}(x, y, z, t) at the magnetometers 26 can be represented by a matrix of influence Q from the Maxwell-constrained coefficient vector (t)=[γ1(t), γ2 (t), . . . γ8 (t)] to the Maxwell-constrained outside magnetic field model {right arrow over (BOUT-MAXWELL)}(x, y, z, t) at the N number of magnetometers 26. Thus, the generic outside magnetic field model {right arrow over (BOUT-MAXWELL)}(x, y, z t) at the N number of magnetometers 26 may be given as:
As briefly discussed above, the processor 30 may parameterize the Maxwell-constrained environmental magnetic field model BENV-PAR based on the measured total residual magnetic field BTOT-MEAS measured by the magnetometers 26 and the known actuated magnetic field BACT-KNOWN at the magnetometers 26 to generate a parameterized environmental magnetic field model BENV-PAR, and in this particular, case, may parameterize the Maxwell-constrained outside magnetic field model {right arrow over (BOUT-MAXWELL)}(x, y, z, t) based on the total residual magnetic field {right arrow over (BTOT-MEAS)}(x, y, z, t) measured by the magnetometers 26 and the known actuated magnetic field {right arrow over (BACT-KNOWN)}(x, y, z, t) at the magnetometers 26 to generate a parameterized outside magnetic field model BOUT-PAR.
In particular, assuming that the very weak MEG magnetic field BMEG can be ignored for purposes of simplicity, it is known that the following equation holds true at each of the magnetometers 26:
where {right arrow over (BTOT)}(x, y, z, t) is the true total magnetic field measurement at the magnetometers 26, {right arrow over (BACT)}(x, y, z, t) is the true actuated magnetic field at the magnetometers 26, and {right arrow over (BOUT)}(x, y, z, t) is the true outside magnetic field at the magnetometers 26.
Substituting the total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) at the magnetometers 26 of the term [1] for the true total residual magnetic field {right arrow over (BTOT)}(x, y, z, t) at the magnetometers 26 of equation [13], the known actuated magnetic field {right arrow over (BACT-KNOWN)}(x, y, z, t) at the magnetometers 26 of equation [5] for the true actuated magnetic field {right arrow over (BACT)}(x, y, z, t) at the magnetometers 26 of equation [13], and the Maxwell-constrained outside magnetic field model {right arrow over (BOUT-MAXWELL)}(x, y, z, t) at the magnetometers 26 of equation [12] for the true outside magnetic field {right arrow over (BOUT)}(x, y, z, t) at the magnetometers 26 of equation [13] yields:
where δ is unknown measurement noise for each magnetometer 26.
The processor 30 may employ any suitable fitting optimization technique (including linear and nonlinear methods, gradient descent, matrix methods, system identification, or machine learning methods, etc.) to fit the Maxwell-constrained coefficient vector (t) of the Maxwell-constrained outside magnetic field model {right arrow over (BOUT-MAXWELL)}(x, y, z, t) to the difference between the total residual magnetic field {right arrow over (BTOT-MEAS)}(x, y, z, t) measured by the magnetometers 26 and the known actuated magnetic field {right arrow over (BACT-KNOWN)}(x, y, z, t) at the magnetometers 26. In the illustrated embodiment, the processor 30 employs a least squares or weighted least squares optimization technique, which serves to minimize the error between collected and known data and estimated data, to accurately estimate the values of the Maxwell-constrained coefficient vectors (t) of the Maxwell-constrained outside magnetic field model {right arrow over (BOUT-MAXWELL)}(x, y, z, t) at the magnetometers 26. That is, the solution that minimizes the difference between the total residual magnetic field {right arrow over (BTOT-MEAS)}(x, y, z, t) measured by each of the magnetometers 26 and the product of the matrix of influence R at the magnetometers 26 and the vector of actuation strengths (t) of the set of magnetic field actuators 28 yields an estimate of the Maxwell-constrained coefficient vector (t) of the Maxwell-constrained outside magnetic field model {right arrow over (BOUT-MAXWELL)}(x, y, z, t) at the magnetometers 26.
Specifically, the least squares estimate of the Maxwell-constrained coefficient vector {right arrow over (γ*)}(t) of the Maxwell-constrained outside magnetic field model {right arrow over (BOUT-MAXWELL)}(x, y, z, t) can be provided as:
{right arrow over (γ*)}(t)=[QTQ]−1QT(BTOT-MEAS(t)−R*(t)), [15]
where Q is the matrix of influence from the Maxwell-constrained coefficient vector (t)=[γ1(t), γ2 (t), . . . γ8(t)] to the Maxwell-constrained outside magnetic field model {right arrow over (BOUT-MAXWELL)}(x, y, z, t) at the N number of magnetometers 26; BTOT-MEAS(t) is the time-varying matrix of total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) at the N number of magnetometers 26; (t) is the actuation strength vector; R is the matrix of influence from the actuation strength vector (t) to the known actuated magnetic field {right arrow over (BACT-KNOWN)}(x, y, z, t) at the N number of magnetometers 26; the superscript T denotes the matrix transpose; and the superscript −1 denotes matrix inversion.
A parameterized outside magnetic field model {right arrow over (BOUT-PAR)}(x, y, z, t) may be generated by substituting the solved Maxwell-constrained coefficient vector {right arrow over (γ*)}(t) into equation [6]. It should be appreciated that the foregoing method transforms a discrete set of the total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) into continuous parameterizations of the outside magnetic field {right arrow over (BOUT)}(x, y, z, t), i.e., the parameterized outside magnetic field model {right arrow over (BOUT-PAR)}(x, y, z, t). This enables the processor 30 to estimate the outside magnetic field BOUT at arbitrary locations in the vicinity from which the measurements of the total residual magnetic field BTOT-MEAS were acquired, i.e., in the vicinity of the signal acquisition unit 18.
As briefly discussed above, the processor 30 may determine the environmental magnetic field estimates BENV-EST at the magnetometers 26 based on the parameterized environmental magnetic field model BENV-PAR, and in this particular case, may determine the outside magnetic field estimates {right arrow over (BOUT-EST)}(x, y, z, t) at the magnetometers 26 based on the parameterized outside magnetic field model {right arrow over (BOUT-PAR)}(x, y, z, t).
In particular, the outside magnetic field estimates {right arrow over (BOUT-EST)}(x, y, z, t) at the magnetometers 26 may be determined by substituting the (x,y,z) locations of the magnetometers 26 into the parameterized outside magnetic field model {right arrow over (BOUT-PAR)}(x, y, z, t); i.e., the outside magnetic field estimates {right arrow over (BOUT-EST)}(x, y, z, t) at the magnetometers 26 may be recovered from the product of the influence matrix Q and the least squares fit values of the Maxwell-constrained coefficient vector {right arrow over (γ*)}(t).
As briefly discussed above, the processor 30 may determine the total residual magnetic field estimates BTOT-EST at the magnetometers 26 based on the known actuated magnetic field BACT-KNOWN at the magnetometers 26 and the environmental magnetic field estimates BOUT-EST and in this particular case, may determine the total residual magnetic field estimates {right arrow over (BTOT-EST)}(x, y, z, t) at magnetometers 26 based on the known actuated magnetic field {right arrow over (BACT-KNOWN)}(x, y, z, t) at the magnetometers 26 and the outside magnetic field estimates {right arrow over (BOUT-EST)}(x, y, z, t), at the magnetometers 26 by summing the known actuated magnetic field {right arrow over (BACT-KNOWN)}(x, y, z, t) at the magnetometers 26 and the outside magnetic field estimates {right arrow over (BOUT-EST)}(x, y, z, t) at the magnetometers 26.
In particular, substituting the total residual magnetic field estimates {right arrow over (BTOT-EST)}(x, y, z, t) at the magnetometers 26 for the true total residual magnetic field {right arrow over (BTOT)}(x, y, z, t) at the magnetometers 26 of equation [13], the known actuated magnetic field {right arrow over (BACT-KNOWN)}(x, y, z, t) at the magnetometers 26 of equation [5] for the true actuated magnetic field {right arrow over (BACT)}(x, y, z, t) at the magnetometers 26 of equation [13], and the outside magnetic field estimates {right arrow over (BOUT-EST)}(x, y, z, t) at the magnetometers 26 for the true outside magnetic field {right arrow over (BOUT)}(x, y, z, t) at the magnetometers 26 of equation [13] yields:
Thus, intuitively, equation [16] need only be solved to accurately infer the total residual magnetic field estimates BTOT-EST at the magnetometers 26.
It can be appreciated that inferring the total residual magnetic field estimates {right arrow over (BTOT-EST)}(x, y, z, t) at that magnetometers 26 based on the total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) taken by all of the magnetometers 26 (including the magnetometer 26 for which the total residual magnetic field estimates {right arrow over (BTOT-EST)}(x, y, z, t) is being inferred) provides a more accurate assessment of the true total residual magnetic field {right arrow over (BTOT)}(x, y, z, t) at each magnetometer 26 than each magnetometer 26 can measure alone, because such inference technique averages out the unknown measurement noise δ of the magnetometers 26 in a rigorous manner. Thus, in effect, the total residual magnetic field measurement {right arrow over (BTOT-MEAS)}(x, y, z, t) taken by each magnetometer 26 is corrected by this inference technique.
Further, as a result of generating a Maxwell-constrained outside magnetic field model BOUT-MAXWELL by applying Maxwell's equations to the generic outside magnetic field model BOUT-MOD, the non-physical portion of the total residual magnetic field measurement {right arrow over (BTOT-MEAS)}(x, y, z, t) taken by each magnetometer 26, and in this particular case, the outside magnetic field BOUT component of the total residual magnetic field measurement {right arrow over (BTOT-MEAS)}(x, y, z, t), is eliminated in the total residual magnetic field estimates {right arrow over (BTOT-EST)}(x, y, z, t) at that magnetometers 26, or at the least, the non-physical portion in the total residual magnetic field estimates {right arrow over (BTOT-EST)}(x, y, z, t) at the magnetometers 26 will be substantially less than the non-physical portion in the total residual magnetic field measurements {right arrow over (BTOT-EST)}(x, y, z, t) at that magnetometers 26. As a result, the processor 30 may control the actuated magnetic field {right arrow over (BACT)}(x, y, z, t) in a manner that more accurately cancels the outside magnetic field {right arrow over (BOUT)}(x, y, z, t), such that the total residual magnetic field {right arrow over (BTOT)}(x, y, z, t) at the magnetometers 26 can be more effectively suppressed.
While acquiring total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) from the magnetometers 26 and inferring the total residual magnetic field estimates {right arrow over (BTOT-EST)}(x, y, z, t) at the magnetometers 26 can be conducted over one time or over all available time, it is preferred that it be conducted over a time window that is updated in time (e.g. from current time t back till time t−T, where T is the time window period and can be constant or variable), since doing so over a longer time period allows the unknown measurement noise δ of the magnetometers 26 to be averaged out.
In one embodiment, the gain and offset of each of the coarse magnetometers 26a can be estimated by comparing the more accurate total residual magnetic field estimate {right arrow over (BTOT-EST)}(x, y, z, t) at each coarse magnetometer 26a that has been inferred from total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) at many magnetometers 26 (including the much more accurate fine magnetometers 26b) to the total residual magnetic field {right arrow over (BTOT-MEAS)}(x, y, z, t) measured by the respective coarse magnetometer 26a.
In another embodiment, a weighted least squares estimate, instead of an unweighted least squares estimate, of the Maxwell-constrained coefficient vector {right arrow over (γ*)}(t) of the Maxwell-constrained outside magnetic field model {right arrow over (BOUT-MAXWELL)}(x, y, z, t) is employed. For example, total residual magnetic field measurements {right arrow over (BTOT-MEASF)}(x, y, z, t) acquired from fine magnetometers 26b are substantially more accurate than total residual magnetic field measurements {right arrow over (BTOT-MEASC)}(x, y, z, t) acquired from coarse magnetometers 26a. Furthermore, due to drifts in the offset or gain of coarse magnetometers 26a over time, newer total residual magnetic field measurements {right arrow over (BTOT-MEASC)}(x, y, z, t) acquired from coarse magnetometers 26b, absent re-calibration, are more accurate than older total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) acquired from the same coarse magnetometers 26b.
As discussed above, acquiring total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) from the magnetometers 26 and inferring the total residual magnetic field estimates {right arrow over (BTOT-EST)}(x, y, z, t) at the magnetometers 26 is preferably conductive over a time window that is updated in time (e.g. from current time t back till time t−T, where T is the time window period and can be constant or variable), since doing so over a longer time period allows the unknown measurement noise δ of the magnetometers 26 to be averaged out.
Thus, in a preferred embodiment, total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) are acquired from the magnetometers 26 and the total residual magnetic field estimates {right arrow over (BTOT-EST)}(x, y, z, t) are inferred at the magnetometers 26 over an updated time window, the total residual magnetic field measurements {right arrow over (BTOT-MEASF)}(x, y, z, t) acquired from fine magnetometers 26b are weighed higher than total residual magnetic field measurements {right arrow over (BTOT-MEASC)}(x, y, z, t) acquired from coarse magnetometers 26a, and newer total residual magnetic field measurements {right arrow over (BTOT-MEASF)}(x, y, z, t) acquired from magnetometers 26 are weighted higher than older total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) acquired from magnetometers 26.
Such weighting can be incorporated into a weighting matrix W that operates on all the total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) acquired from magnetometers 26. Elements of the weighting matrix W can be selected to be inversely proportional to the measurement variance of each magnetometer 26, such that total residual magnetic field measurements {right arrow over (BTOT-MEASC)}(x, y, z, t) acquired from coarse magnetometers 26a (which have a high measurement variance (and thus relatively low accuracy) have a small weight, while total residual magnetic field measurements {right arrow over (BTOT-MEASF)}(x, y, z, t) acquired from fine magnetometers 26b (which have a low measurement variance (and thus relatively high accuracy) have a large weight. Furthermore, elements of the weighting matrix W may decrease as the total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) acquired from magnetometers 26 age, such that older total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) have a small weight and newer total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) have a large weight. This decrease in the elements of the weighting matrix W may be linear, quadratic, stepwise, or have some other functional dependence that can be selected by intuition or by mathematical optimization by methods known to those of ordinary skill in the art of optimization, control and signal processing, or system identification. In one embodiment, the functional dependence may match the time scale of how quickly the gain or offset of the coarse magnetometers drift in time.
The unweighted least squares estimate of equation [15] can then be modified with the weighting matrix W as:
{right arrow over (γ*)}((t−T)→t)=[QTWTWQ]−1QTWTW(BTOT-MEAS((t−T)→t)−R*((t−T)→t)), [17]
where Q is the matrix of influence from the Maxwell-constrained coefficient vector (t)=[γ1(t), γ2(t), . . . γ8 (t)] to the Maxwell-constrained outside magnetic field model {right arrow over (BOUT-MAXWELL)}(x, y, z, t) at the N number of magnetometers 26; ((t−T)→t) is an updated time window; BTOT-MEAS((t−T)→t) is the time-varying matrix of the total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) at the N number of magnetometers 26 over the time window ((t−T)→t); ((t−T)→t) is the time-varying vector of actuation strengths over the time window ((t−T)→t); R is the matrix of influence from the actuation strength vector ((t−T)→t) to the known actuated magnetic field {right arrow over (BACT-KNOWN)}(x, y, z, t) at the N number of magnetometers 26; the superscript T denotes the matrix transpose; the superscript −1 denotes matrix inversion; and W is the weighting matrix.
Although all three of the directional components of the outside magnetic field BOUT and actuated magnetic field BACT, and thus all three of the directional components of the total residual magnetic field BTOT, have been considered when inferring the total residual magnetic field estimates BTOT-EST at the magnetometers 26, it should be appreciated that only one or two directional components of the outside magnetic field BOUT and/or actuated magnetic field BACT may be considered when inferring the total residual magnetic field estimates BTOT-EST at the magnetometers 26. Furthermore, although all three directional components of the total residual magnetic field BTOT-MEAS have been described as being measured and estimated at the same location or virtually at the same location for each magnetometer 26, less than three directional components of the total residual magnetic field BTOT-MEAS may be measured and/or estimated at the same location or virtually at the same location for each magnetometer 26.
As discussed above, Maxwell's equations can be applied to the total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) to eliminate, or at least substantially reduce, the non-physical portion of the MEG magnetic field {right arrow over (BMEG)}(x, y, z, t) component of the total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) such that the MEG magnetic field {right arrow over (BMEG)}(x, y, z, t) may be more accurately determined. The non-physical portion of the MEG magnetic field {right arrow over (BMEG)}(x, y, z, t) component of the total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) can be reduced by performing the same procedure used to reduce the non-physical portion of the MEG magnetic field {right arrow over (BMEG)}(x, y, z, t) component of the total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) described above, with the exception that a generic model of the environmental magnetic field {right arrow over (BENV-MOD)}(x, y, z, t) containing both initial basis functions (e.g., the 0th order and 1st order basis functions) for the {right arrow over (BOUT)}(x, y, z, t) and initial basis functions for the MEG magnetic field {right arrow over (BMEG)}(x, y, z, t).
Thus, the processor 30 may be configured for inferring total residual magnetic field estimates {right arrow over (BTOT-EST)}(x, y, z, t) at the magnetometers 26 by (1) acquiring the measurements of the total residual magnetic field {right arrow over (BTOT-MEAS)}(x, y, z, t) from the magnetometers 26, as exemplified by the previous discussion related to equations [1] and [2]; (2) determining the known actuated magnetic field {right arrow over (BACT-KNOWN)}(x, y, z, t) at the magnetometers 26 based on a known profile of the set of magnetic field actuators 28 and the actuation strengths of the magnetic field actuators 28, as exemplified by previous discussion relating to equations [3]-[5]; (3) generating a generic model of the environmental magnetic field {right arrow over (BENV-MOD)}(x, y, z, t) in the vicinity of the magnetometers 26 in a similar manner that the generic outside magnetic model {right arrow over (BOUT-MOD)}(x, y, z, t) is generated in the discussion related to equation [6] with the exception that the generic environmental magnetic field model {right arrow over (BENV-MOD)}(x, y, z, t) comprises additional initial basis functions for the MEG magnetic field {right arrow over (BMEG)}(x, y, z, t); (4) constraining the environmental magnetic field model {right arrow over (BENV-MOD)}(x, y, z, t) to generate a Maxwell-constrained model of the environmental magnetic field {right arrow over (BENV-MAXWELL)}(x, y, z, t) that satisfies Maxwell's equations, in a similar manner that generic outside magnetic model {right arrow over (BOUT-MOD)}(x, y, z, t) is constrained in the discussion related to equations [7]-[12], with the exception that the initial basis functions for the outside magnetic field {right arrow over (BOUT-MOD)}(x, y, z, t) and the MEG magnetic field {right arrow over (BMEG)}(x, y, z, t) are collectively reduced; (5) parameterizing the Maxwell-constrained environmental magnetic field model {right arrow over (BENV-MAXWELL)}(x, y, z, t) based on the measured total residual magnetic field {right arrow over (BTOT-MEAS)}(x, y, z, t) measured by the magnetometers 26 and the known actuated magnetic field {right arrow over (BACT-KNOWN)}(x, y, z, t) at the magnetometers 26 to generate a parameterized environmental magnetic field model {right arrow over (BENV-ENV)}(x, y, z, t), in a similar manner that the {right arrow over (BOUT-MAXWELL)}(x, y, z, t) is constrained in the discussion related to equations [13]-[15]; (6) determining the environmental magnetic field estimates {right arrow over (BENV-EST)}(x, y, z, t) at the magnetometers 26 based on the parameterized environmental magnetic field model {right arrow over (BENV-ENV)}(x, y, z, t) by substituting the locations of the magnetometers 26 into the parameterized environmental magnetic field model {right arrow over (BENV-ENV)}(x, y, z, t); and (7) determining the total residual magnetic field estimates {right arrow over (BTOT-EST)}(x, y, z, t) at the magnetometers 26 based on the known actuated magnetic field {right arrow over (BACT-KNOWN)}(x, y, z, t) at the magnetometers 26 and the environmental magnetic field estimates {right arrow over (BENV-EST)}(x, y, z, t) at the magnetometers 26, in the same manner as the total residual magnetic field estimates {right arrow over (BTOT-EST)}(x, y, z, t) is determined in the discussion related to equation [16].
In another embodiment, the processor 30 may be configured for more accurately estimating a magnetic field component of measurements of an arbitrary magnetic field BARB-MEAS at the magnetometers 26b by (1) generating a generic magnetic field model BARB-MOD of a plurality of magnetic field components of the arbitrary magnetic field BARB in the vicinity of the magnetometers 26, with the generic magnetic field model BARB-MOD comprising a plurality of basis functions having multiple sets of basis functions respectively corresponding to a plurality of magnetic components of the arbitrary magnetic field measurements BARB-MEAS at the magnetometers 26; (2) parameterizing the generic magnetic field model BOUT-MOD by simultaneously fitting coefficients of the plurality of basis functions at least partially based on the arbitrary magnetic field measurements BARB-MEAS at the magnetometers 26, thereby yielding a parameterized magnetic field model BARB-PAR of the magnetic field components of the arbitrary magnetic field BARB in the vicinity of the magnetometers 26; and estimating one of the magnetic field components of the arbitrary magnetic field BARB at each of the fine magnetometers 26b based on one of the multiple sets of basis functions of the parameterized magnetic field model BARB-PAR, and optionally estimating additional ones of the magnetic field components of the arbitrary magnetic field BARB at each of the fine magnetometers 26b based additional ones of the multiple sets of basis functions of the parameterized magnetic field model BARB-PAR.
In particular, when the N number of total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) respectively taken by an N number of magnetometers 26 is greater than a p number of basis functions (i.e., modes) in an arbitrary magnetic field {right arrow over (BARB)}(x, y, z, t) (i.e., p<N), a generic model of the arbitrary magnetic field {right arrow over (BARB-MOD)}(x, y, z, t) containing basis functions corresponding to modes of different magnetic field components in the arbitrary magnetic field {right arrow over (BARB)}(x, y, z, t) may be generated. The generic arbitrary magnetic field model {right arrow over (BARB-MOD)}(x, y, z, t) may be represented as an influence matrix Z from a coefficient vector (t) containing p number of coefficients [γ1(t), γ2 (t), . . . γp (t)] to the arbitrary magnetic field {right arrow over (BARB)}(x, y, z, t) at the N number of magnetometers 26. The influence matrix Z has an N number of row vectors and a p number of column vectors, as follows:
The N number of row vectors correspond to the N number of total residual magnetic field measurements {right arrow over (BTOT-MEAS)}(x, y, z, t) respectively taken by the N number of magnetometers 26, while the p number of column vectors respectively correspond to the basis functions (i.e., modes) of the arbitrary magnetic field {right arrow over (BARB)}(x, y, z, t). Significantly, the influence matrix Z contains multiple influence matrices QRETAIN, Q′DISCARD, Q″DISCARD, . . . , as follows:
Z=[QRETAIN Q′DISCARD Q″DISCARD . . . ], [19]
where the column vectors of the influence matrix QRETAIN respectively correspond to the modes of the arbitrary magnetic field {right arrow over (BARB)} (x, y, z, t) to be retained; the column vectors of the influence matrix Q′DISCARD respectively correspond to the modes of the arbitrary magnetic field {right arrow over (BARB)}(x, y, z, t) to be discarded; the column vectors of the influence matrix Q′DISCARD respectively correspond to additional modes of the arbitrary magnetic field {right arrow over (BARB)}(x, y, z, t) to be discarded; and so forth. The column vectors of the influence matrix Q′DISCARD are orthogonal to the column vectors of the influence matrix QRETAIN, the column vectors of the influence matrix Q″DISCARD are orthogonal to the column vectors of the influence matrices QRETAIN and QDISCARD, and so forth. Thus, the influence matrix Z is defined by a concatenation of orthogonal influence matrices QRETAIN, Q′DISCARD, Q″DISCARD, . . . , with the column vectors to the right of the influence matrix QRETAIN being considered the rejection space, with the basis functions in the influence.
Although the influence matrix Z is illustrated here as being concatenated with only one influence matrix QRETAIN to be retained and several influence matrices QDISCARD to be discarded, it should be appreciated that the influence matrix Z may be concatenated with multiple influence matrices QRETAIN to be retained with or without one or more influence matrices QDISCARD to be discarded. Thus, the influence matrix Z may be concatenated with any combination of influence matrices QRETAIN to be retained and/or influence matrices QDISCARD to be discarded as long the concatenated influence matrices QRETAIN and/or QDISCARD contain mutually exclusive modes of multiple magnetic field components.
The generic arbitrary magnetic field model {right arrow over (BARB-MOD)}(x, y, z, t) may then be parameterized to generate a parameterized model of the arbitrary magnetic field {right arrow over (BARB-PAR)}(x, y, z) by determining the least squares estimate of the coefficient vector {right arrow over (γ*)}(t) in the manner discussed above with respect to equation [15] with the exception that the influence matrix Q is replaced with the concatenated influence matrix Z, as follows:
{right arrow over (γRETAIN*)}(t)[ZTZ]−1ZT({right arrow over (BTOT-MEAS)}(x,y,z,t)−R*(t)){1:pRETAIN}; [20a]
{right arrow over (γ′DISCARD*)}(t)=[ZTZ]−1ZT({right arrow over (BTOT-MEAS)}(x,y,z,t)−R*(t)){pRETAIN+1:pRETAIN+p′DISCARD}; [20b]
{right arrow over (y″DISCARD*)}(t)=[ZTZ]−1ZT({right arrow over (BTOT-MEAS)}(x,y,z,t)−R*(t){pRETAIN+p′DISCARD:pRETAIN+p′DISCARD+p″DISCARD}, and so forth, [20c]
where BTOT-MEAS(x, y, z, t), Z, R, and (t) have been defined above; the notation X{A:B} means take the Ath through Bth elements of X; {right arrow over (γRETAIN*)}(t) is the least squares solution of the coefficient vector corresponding to the influence matrix Q to be retained; pRETAIN are the number of modes of the arbitrary magnetic field {right arrow over (BARB)}(x, y, z, t) corresponding to the influence matrix Q to be retained; {right arrow over (γ′DISCARD*)}(t) is the least squares solution of the coefficient vector corresponding to the influence matrix Q′ to be discarded; p′DISCARD is the number of modes of the arbitrary magnetic field {right arrow over (BARB)}(x, y, z, t) corresponding to the influence matrix Q′ to be discarded {right arrow over (γ″DISCARD*)}(t) is the least squares solution of the coefficient vector corresponding to the influence matrix Q″ to be discarded; p′DISCARD is the number of modes of the arbitrary magnetic field {right arrow over (BARB)} (x, y, z, t) corresponding to the influence matrix Q″ to be discarded, and so forth.
Thus, it can be appreciated that the promotion of a single influence matrix Q to be retained to an influence matrix Z containing both an influence matrix Q to be retained and influence matrices Q′, Q″ . . . to be discarded, in effect, fusing modes of the arbitrary magnetic field {right arrow over (BARB)} (x, y, z, t) derived from multiple models of the arbitrary magnetic field {right arrow over (BARB)}(x, y, z, t), enables separation and precise specification of the modes of the arbitrary magnetic field {right arrow over (BARB)}(x, y, z, t) to be retained and to be discarded. Thus, by simultaneously fitting the coefficient vector {right arrow over (γRETAIN*)}(t) and coefficient vectors {right arrow over (γ′DISCARD*)}(t) and {right arrow over (γ″DISCARD*)}(t) to the difference between the total residual magnetic field {right arrow over (BTOT-MEAS)}(x, y, z, t) measured by the magnetometers 26 and the known actuated magnetic field {right arrow over (BACT-KNOWN)}(x, y, z, t) at the magnetometers 26, the accuracy of the solution for the coefficient vector {right arrow over (γRETAIN*)}(t) is increased.
Arbitrary magnetic field estimates {right arrow over (BARB-EST)}(x, y, z, t) may be determined at the fine magnetometers 26b for any particular magnetic field component of interest by substituting the (x,y,z) locations of the fine magnetometers 26b into the basis functions of the parameterized arbitrary magnetic field model {right arrow over (BARB-PAR)}(x, y, z, t) corresponding to that magnetic field component of interest; i.e., such magnetic field component of the arbitrary magnetic field estimates BARB-EST(x, y, z, t) at the fine magnetometers 26b may be recovered from the product of the influence matrix Z and the least squares fit values of the coefficient vector {right arrow over (γ*)}(t) corresponding to that magnetic field component. For example, one magnetic field component of the arbitrary magnetic field estimates {right arrow over (BARB-EST)}(x, y, z, t) at the fine magnetometers 26 may be recovered from the product of the influence matrix Z and the least squares fit values of the coefficient vector {right arrow over (γRETAIN*)}(t) to be retained. The magnetic field components of the arbitrary magnetic field estimates {right arrow over (BARB-EST)}(x, y, z, t) that are not of interest may simply be ignored, and therefore, not estimated at the fine magnetometers 26b.
In one specific embodiment, the processor 30 may employ the equations [19] and [20a]-[20c] to distinguish between a physical outside magnetic field {right arrow over (BOUT-P)}(x, y, z, t) component and a non-physical outside magnetic field {right arrow over (BOUT-NP)}(x, y, z, t) component of the total residual magnetic field measurements BTOT(x, y, z, t) acquired from the magnetometers 26. In particular, based on equation [6] above, the generic outside magnetic field model BOUT-MOD (x, y, z, t) can be partitioned into a physically possible magnetic field model {right arrow over (BOUT-P-MOD)}(x, y, z, t) that satisfies Maxwell's equations (corresponded to the Maxwell-constrained outside magnetic field model {right arrow over (BOUT-MAXWELL)}(x, y, z, t)), and a physically impossible magnetic field {right arrow over (BOUT-NP-MOD)}(x, y, z, t) that does not satisfy Maxwell's equations. In this case, the outside magnetic field {right arrow over (BOUT-MOD)}(x, y, z, t) has twelve basis functions (i.e., modes), and in particular, αx(t), αxx(t)x, αxy(t)y, αxz(t)z, αy(t), αyx(t)x, αyy(t)y, αyz(t)z, αz(t), αzx(t)x, αzy(t)y, αzz(t)z) and a coefficient vector [γ1(t), γ2(t), . . . γ12(t)] (i.e., p=12). The physically possible magnetic field model {right arrow over (BOUT-P-MOD)}(x, y, z, t) has eight twelve basis functions (i.e., modes) and a coefficient vector [γ1(t), γ2(t), . . . γ8(t)], while physically possible magnetic field model {right arrow over (BOUT-P-MOD)}(x, y, z, t) has four twelve basis functions (i.e., modes) and a coefficient vector [γ9(t), γ2(t), . . . γ12(t)].
The modes of the physically possible magnetic field model {right arrow over (BOUT-P-MOD)}(x, y, z, t) are to be retained, whereas the modes of the physically impossible magnetic field model {right arrow over (BOUT-NP-MOD)}(x, y, z, t) are to be discarded. Thus, an influence matrix QOUT-PHYS by a coefficient vector (t) to the physical outside magnetic field model {right arrow over (BOUT-P-MOD)}(x, y, z, t) at the N number of magnetometers 26 can be generated, and an influence matrix QOUT-NP by a coefficient vector (t) to the physically impossible magnetic field model {right arrow over (BOUT-NP-MOD)}(x, y, z, t) at the N number of magnetometers 26 can be generated.
The influence matrix QOUT-P has a size (N×pOUT-P), where pOUT-P is the number of modes in the physically possible magnetic field model {right arrow over (BOUT-P-MOD)}(x, y, z, t) (in this case, pOUT-P=8). The influence matrix QOUT-NP has a size (N×pOUT-NP), where pOUT-NP is the number of modes in the physically impossible magnetic field model {right arrow over (BOUT-NP-MOD)}(x, y, z, t) (in this case, pOUT-NP=4). The influence matrices QOUT-P and QOUT-NP may be concatenated into an influence matrix ZOUT from a coefficient vector (t) containing twelve coefficients [γ1(t), γ2(t), . . . γ12 (t)] (i.e., p=12) to the outside magnetic field {right arrow over (BOUT)}(x, y, z, t) at the N number of magnetometers 26. In this case, the influence matrix ZOUT may take the form of:
Z
OUT=[QOUT-P Q′OUT-NP], [21]
where the (p−4) leftmost column vectors of the influence matrix Z are the column vectors of the influence matrix QOUT-P that respectively correspond to the modes of the generic outside magnetic field model {right arrow over (BOUT-MOD)} (x, y, z, t) to be retained (i.e., the modes of the physically possible magnetic field model {right arrow over (BOUT-P-MOD)}(x, y, z, t)); and 4 rightmost column vectors of the influence matrix ZOUT are the column vectors of the influence matrix Q′OUT-NP that respectively correspond to the modes of the generic outside magnetic field model {right arrow over (BOUT-MOD)}(x, y, z, t) to be discarded (i.e., the modes of the physically impossible magnetic field model {right arrow over (BOUT-NP-MOD)}(x, y, z, t)).
The generic outside magnetic field model {right arrow over (BOUT-MOD)}(x, y, z, t) may then be parameterized to generate a parameterized model of the outside magnetic field {right arrow over (BOUT-PAR)}(x, y, z) by determining the least squares estimate of the coefficient vector {right arrow over (γ*)}(t) in accordance with modified equations [20a] and [20b] as follows:
{right arrow over (γP*)}(t)=[ZTZ]−1ZT({right arrow over (BTOT-MEAS)}(x,y,z,t)−R*(t)){1:pP}; and [22a]
{right arrow over (γNP*)}(t)=[ZTZ]−1ZT({right arrow over (BTOT-MEAS)}(x,y,z,t)−R*(t)){pP+1: pNP}, [22b]
where {right arrow over (BTOT-MEAS)}(x, y, z, t), Z, R, (t), pP, and pNP have been defined above; the notation X{A:B} means take the Ath through Bth elements of X; {right arrow over (γP*)}(t) is the least squares solution of the coefficient vector corresponding to the influence matrix QP respectively corresponding to the modes of the physically possible magnetic field model {right arrow over (BP-MOD)}(x, y, z, t); and {right arrow over (γNP*)}(t) is the least squares solution of the coefficient vector corresponding to the influence matrix Q′NP respectively corresponding to the modes of the physically impossible magnetic field model {right arrow over (BNP-MOD)}(x, y, z, t).
The processor 30 may then estimate the physical outside magnetic field {right arrow over (BOUT-P-EST)}(x, y, z, t) at the fine magnetometers 26b by substituting the (x,y,z) locations of the fine magnetometers 26b into the basis functions of the parameterized outside magnetic field model {right arrow over (BOUT-PAR)}(x, y, z, t) corresponding to the modes of the physical outside magnetic field {right arrow over (BOUT-P)}(x, y, z, t); i.e., the physical outside magnetic field estimates {right arrow over (BOUT-P-EST)}(x, y, z, t) at the fine magnetometers 26b may be recovered from the product of the influence matrix ZOUT and the least squares fit values of the coefficient vector {right arrow over (γP*)}(t) corresponding to the modes of the physical outside magnetic field {right arrow over (BOUT-P)}(x, y, z, t). The processor 30 may then use the physical outside magnetic field estimates {right arrow over (BOUT-P-EST)}(x, y, z, t) at the fine magnetometers 26b to control the set of magnetic field actuators 28 to at least partially cancel the outside magnetic field {right arrow over (BOUT)}(x, y, z, t), thereby suppressing the total residual magnetic field {right arrow over (BTOT)} (x, y, z, t) to the baseline level at the fine magnetometers 26b. The non-physical outside magnetic field {right arrow over (BOUT-P)}(x, y, z, t) at the fine magnetometers 26b may simply be ignored, and therefore, not estimated at the fine magnetometers 26b.
In another specific embodiment, instead of distinguishing between a physical outside magnetic field {right arrow over (BOUT-P)}(x, y, z, t) component and a non-physical outside magnetic field {right arrow over (BOUT-NP)}(x, y, z, t) component of the total residual magnetic field measurements {right arrow over (BTOT)}(x, y, z, t) acquired from the magnetometers 26, the processor 30 may employ the equations [19] and [20a]-[20c] to distinguish the MEG magnetic field BMEG component (i.e., the portion represented by the oval 60 in
In particular, while MEG magnetic field BMEG was ignored in equation [13], the outside magnetic field {right arrow over (BOUT)}(x, y, z, t) in equation [13] can be replaced with the environmental magnetic field {right arrow over (BENV)}(x, y, z, t), which includes the outside magnetic field {right arrow over (BOUT)}(x, y, z, t) and a MEG magnetic field model {right arrow over (BMEG-MOD)}(x, y, z, t). Thus, a generic environmental magnetic field model {right arrow over (BENV-MOD)}(x, y, z, t) may be defined and partitioned into a MEG magnetic field model {right arrow over (BMEG-MOD)}(x, y, z, t) and an outside magnetic field model {right arrow over (BOUT-MOD)}(x, y, z, t), where the modes of the MEG magnetic field model {right arrow over (BMEG-MOD)}(x, y, z, t) are to be retained, and the modes of the generic outside magnetic field model {right arrow over (BOUT-MOD)}(x, y, z, t) are to be discarded.
A matrix of influence QMEG by a coefficient vector {right arrow over (γMEG)}(t) to the MEG magnetic field model {right arrow over (BMEG-MOD)}(x, y, z, t) at the N number of magnetometers 26 can be generated. The matrix of influence QMEG may be generated using mathematical or numerical modeling (e.g., by simulating the MEG magnetic field BMEG emanating from a brain to different spatial locations, e.g., at the magnetometers 26) or by the performance of calibration measurements ahead of time (i.e., measure the actual MEG magnetic field BMEG emanating from a brain at different spatial locations, e.g., at the magnetometers 26).
Similarly, another matrix of influence QOUT by a coefficient vector {right arrow over (γOUT)}(t) to the generic outside magnetic field model {right arrow over (BOUT-MOD)}(x, y, z, t) at the N number of magnetometers 26, can be generated. The matrix of influence QOUT may be generated using mathematical or numerical modeling (e.g., by simulating the outside magnetic field BOUT at different spatial locations, e.g., at the magnetometers 26) or by the performance of calibration measurements ahead of time (i.e., measure the actual outside magnetic field BOUT at different spatial locations, e.g., at the magnetometers 26).
The influence matrices QMEG and QOUT may be generated using a variety of matrix factorization methods, including SVD, the QR, LU, Jordan and other eigenvalue-based decompositions, gradient descent optimization, nonnegative matrix factorization and other types of matrix factorization, and similar methods known to a persons of ordinary skill in the art of signal processing, systemic identification, optimization, control theory, or neuroscience.
The influence matrix QMEG has a size (N×pMEG), where pMEG is the number of modes in the MEG magnetic field model {right arrow over (BMEG-MOD)}(x, y, z, t). The influence matrix QOUT has a size (N×pOUT), where pOUT is the number of modes in the generic outside magnetic field model {right arrow over (BOUT-MOD)}(x, y, z, t). The influence matrices QMEG and QOUT may be concatenated into an influence matrix Z from a coefficient vector (t) to the environmental magnetic field {right arrow over (BENV)}(x, y, z, t) at the N number of magnetometers 26. In this case, the influence matrix Z may take the form of:
Z=[QMEG Q′OUT], [23]
where the column vectors of the influence matrix QMEG respectively correspond to the modes of the environmental magnetic field model {right arrow over (BMEG-ENV)}(x, y, z, t) to be retained (i.e., the modes of the MEG magnetic field model {right arrow over (BMEG-MOD)}(x, y, z, t)); and the column vectors of the influence matrix Q′OUT respectively correspond to the modes of the environmental magnetic field model {right arrow over (BMEG-ENV)}(x, y, z, t) to be discarded (i.e., the modes of the generic outside magnetic field model {right arrow over (BOUT-MOD)}(x, y, z, t)).
The generic environmental magnetic field model {right arrow over (BENV-MOD)}(x, y, z, t) may then be parameterized to generate a parameterized model of the environmental magnetic field {right arrow over (BENV-PAR)}(x, y, z) by determining the least squares estimate of the coefficient vector {right arrow over (γ*)}(t) in accordance with modified equations [20a] and [20b] as follows:
{right arrow over (γMEG*)}(t)=[ZTZ]−1ZT({right arrow over (BTOT-MEAS)}(x,y,z,t)−R*(t)){:pMEG}; and [24a]
{right arrow over (γOUT*)}(t)=[ZTZ]−1ZT({right arrow over (BTOT-MEAS)}(x,y,z,t)−R*(t)){pMEG+1:pOUT}, [24b]
where {right arrow over (BTOT-MEAS)}(x, y, z, t), Z, R, (t), pMEG, and pOUT have been defined above; the notation X{A:B} means take the Ath through Bth elements of X; {right arrow over (γMEG*)}(t) is the least squares solution of the coefficient vector corresponding to the influence matrix QMEG respectively corresponding to the modes of the MEG magnetic field model {right arrow over (BMEG-MOD)}(x, y, z, t); and {right arrow over (γOUT*)}(t) is the least squares solution of the coefficient vector corresponding to the influence matrix QOUT respectively corresponding to the modes of the generic outside magnetic field model {right arrow over (BOUT-MOD)}(x, y, z, t).
The processor 30 may then estimate the MEG magnetic field {right arrow over (BMEG-EST)}(x, y, z, t) at the fine magnetometers 26b by substituting the (x,y,z) locations of the fine magnetometers 26b into the basis functions of the parameterized environmental magnetic field model {right arrow over (BENV-PAR)}(x, y, z, t) corresponding to the modes of the MEG magnetic field {right arrow over (BMEG)}(x, y, z, t); i.e., the MEG magnetic field estimates {right arrow over (BMEG-EST)}(x, y, z, t) at the fine magnetometers 26b may be recovered from the product of the influence matrix Z and the least squares fit values of the coefficient vector {right arrow over (γ*)}(t) corresponding to the modes of the MEG magnetic field {right arrow over (BMEG)}(x, y, z, t). The processor 30 may then derive the MEG signals SMEG from the MEG magnetic field estimates {right arrow over (BMEG-EST)}(x, y, z, t) at the fine magnetometers 26b.
The outside magnetic field {right arrow over (BOUT)}(x, y, z, t) component of the environmental magnetic field estimates {right arrow over (BENV-EST)}(x, y, z, t) may simply be ignored, and therefore, not estimated at the fine magnetometers 26b. Alternatively, the processor 30 may estimate the outside magnetic field {right arrow over (BOUT-EST)}(x, y, z, t) at the fine magnetometers 26b by substituting the (x,y,z) locations of the fine magnetometers 26b into the basis functions of the parameterized environmental magnetic field model {right arrow over (BENV-PAR)}(x, y, z, t) corresponding to the modes of the outside magnetic field {right arrow over (BOUT)}(x, y, z, t); i.e., the outside magnetic field estimates {right arrow over (BOUT-EST)}(x, y, z, t) at the fine magnetometers 26b may be recovered from the product of the influence matrix Z and the least squares fit values of the coefficient vector {right arrow over (γ*)}(t) corresponding to the modes of the outside magnetic field {right arrow over (BOUT-EST)}(x, y, z, t). The processor 30 may then use the outside magnetic field estimates {right arrow over (BOUT-EST)}(x, y, z, t) to control the set of magnetic field actuators 28 to at least partially cancel the outside magnetic field {right arrow over (BOUT)}(x, y, z, t), thereby suppressing the total residual magnetic field {right arrow over (BTOT)}(x, y, z, t) to the baseline level at the fine magnetometers 26b.
In another specific embodiment, instead of distinguishing between the MEG magnetic field BMEG component and the outside magnetic field BOUT component of the total residual magnetic field measurements {right arrow over (BTOT)}(x, y, z, t) acquired from the magnetometers 26, the processor 30 may employ the equations [19] and [20a]-[20c] to further distinguish between a MEG magnetic field {right arrow over (BMEG-OI)}(x, y, z, t) component of interest and a MEG magnetic field {right arrow over (BMEG-NOI)}(x, y, z, t) component not of interest of the measured total residual magnetic field BTOT-MEAS acquired from the magnetometers 26. For example, a portion of the MEG magnetic field {right arrow over (BMEG)}(x, y, z, t) that is generated by neural activity in the right temporal lobe of the brain 14 may be of interest, whereas the remaining portion of the MEG magnetic field {right arrow over (BMEG)}(x, y, z, t) that is generated by neural activity in other regions of the brain 14 may not be of interest.
Thus, a generic environmental magnetic field model {right arrow over (BENV-MOD)}(x, y, z, t) may be defined and partitioned into an outside magnetic field model {right arrow over (BOUT-MOD)}(x, y, z, t) and a MEG magnetic field model {right arrow over (BMEG-MOD)}(x, y, z, t), which may be further partitioned into a MEG magnetic field model of interest {right arrow over (BMEG-OI-MOD)}(x, y, z, t) and a MEG magnetic field model not of interest BMEG-NOI-MOD(x, y, z, t), where the modes of the MEG magnetic field model of interest {right arrow over (BMEG-OI-MOD)}(x, y, z, t) are to be retained, and the modes of the generic outside magnetic field model {right arrow over (BOUT-MOD)}(x, y, z, t) and the MEG magnetic field model not of interest BMEG-NOI-MOD(x, y, z, t) are to be discarded.
The matrix of influence QOUT may generated in the manner described above, whereas a matrix of influence QMEG-OI by a coefficient vector (t) to the MEG magnetic field model of interest {right arrow over (BMEG-OI-MOD)}(x, y, z, t) at the N number of magnetometers 26 can be generated, and a matrix of influence QMEG-NOI by a coefficient vector (t) to the MEG magnetic field model of not of interest {right arrow over (BMEG-NOI-MOD)}(x, y, z, t) at the N number of magnetometers 26 can be generated. The matrices of influence QMEG-OI and QMEG-NOI may be generated using mathematical or numerical modeling (e.g., by simulating the MEG magnetic field of interest BMEG-OI or the MEG magnetic field not of interest BMEG-NOI emanating from a brain to different spatial locations, e.g., at the magnetometers 26) or by the performance of calibration measurements ahead of time (i.e., measure the actual MEG magnetic field of interest BMEG-OI or the MEG magnetic field not of interest BMEG-NOI emanating from a brain at different spatial locations, e.g., at the magnetometers 26).
The influence matrices QMEG-OI and QMEG-NOI may be generated using a variety of matrix factorization methods, including SVD, the QR, LU, Jordan and other eigenvalue-based decompositions, gradient descent optimization, nonnegative matrix factorization and other types of matrix factorization, and similar methods known to a persons of ordinary skill in the art of signal processing, systemic identification, optimization, control theory, or neuroscience.
The influence matrix QMEG-OI has a size (N×pMEG-OI), where pMEG-OI is the number of modes in the MEG magnetic field model of interest {right arrow over (BMEG-OI-MOD)}(x, y, z, t). The influence matrix QMEG-NOI has a size (N×pMEG-NOI), where pMEG-NOI is the number of modes in the MEG magnetic field model not of interest {right arrow over (BMEG-NOI-MOD)}(x, y, z, t). The influence matrices QMEG-OI, QMEG-NOI, and QOUT may be concatenated into an influence matrix Z from a coefficient vector (t) to the environmental magnetic field {right arrow over (BENV)}(x, y, z, t) at the N number of magnetometers 26. In this case, the influence matrix Z may take the form of:
Z=[QMEG-OI QMEG-NOI QOUT], [25]
where the column vectors of the influence matrix QMEG-OI respectively correspond to the modes of the environmental magnetic field model {right arrow over (BMEG-ENV)}(x, y, z, t) to be retained (i.e., the modes of the MEG magnetic field model of interest {right arrow over (BMEG-0I-MOD)}(x, y, z, t)); the column vectors of the influence matrix QMEG-NOI respectively correspond to the modes of the environmental magnetic field model {right arrow over (BMEG-ENV)}(x, y, z, t) to be discarded (i.e., the modes of the MEG magnetic field model not of interest {right arrow over (BMEG-NOI-MOD)}(x, y, z, t)); and column vectors of the influence matrix QOUT respectively correspond to the modes of the environmental magnetic field model {right arrow over (BMEG-ENV)}(x, y, z, t) to be discarded (i.e., the modes of the generic outside magnetic field model {right arrow over (BOUT-MOD)}(x, y, z, t)).
The generic environmental magnetic field model {right arrow over (BENV-MOD)}(x, y, z, t) may then be parameterized to generate a parameterized model of the magnetic field {right arrow over (BOUT-PAR)}(x, y, z) by determining the least squares estimate of the coefficient vector {right arrow over (γ*)}(t) in accordance with modified equations [20a]-[20c] as follows:
{right arrow over (γMEG-OI*)}(t)=[ZTZ]−1ZT({right arrow over (BTOT-MEAS)}(x,y,z,t)−R*(t)){:pMEG-OI}; [26a]
{right arrow over (γMEG-NOI*)}(t)=[ZTZ]−1ZT({right arrow over (BTOT-MEAS)}(x,y,z,t)−R*(t)){pMEG-OI+1:pMEG-OI+pMEG-NOI}; and [26b]
{right arrow over (γOUT*)}(t)=[ZTZ]−1ZT({right arrow over (BTOT-MEAS)}(x,y,z,t)−R*(t)){pMEG-OI+pMEG-NOI+1:pMEG-OI+pMEG-NOI+pOUT}, [26c]
where {right arrow over (BTOT-MEAS)}(x, y, z, t), Z, R, (t), pMEG-OI, pMEG-NOI, and pOUT have been defined above; the notation X{A:B} means take the Ath through Bth elements of X; {right arrow over (γMEG-OI*)}(t) is the least squares solution of the coefficient vector corresponding to the influence matrix QMEG-GI respectively corresponding to the modes of the MEG magnetic field model {right arrow over (BMEG-OI-MOD)}(x, y, z, t); {right arrow over (γMEG-NOI*)}(t) is the least squares solution of the coefficient vector corresponding to the influence matrix QMEG-NOI respectively corresponding to the modes of the MEG magnetic field model not of interest {right arrow over (BMEG-NOI-MOD)}(x, y, z, t).
The processor 30 may then estimate the MEG magnetic field of interest {right arrow over (BMEG-OI-EST)}(x, y, z, t) at the fine magnetometers 26b by substituting the (x,y,z) locations of the fine magnetometers 26b into the basis functions of the parameterized environmental magnetic field model {right arrow over (BENV-PAR)}(x, y, z, t) corresponding to the modes of the MEG magnetic field of interest {right arrow over (BMEG-OI)}(x, y, z, t); i.e., the MEG magnetic field of interest estimates {right arrow over (BMEG-OI-EST)}(x, y, z, t) at the fine magnetometers 26b may be recovered from the product of the influence matrix Z and the least squares fit values of the coefficient vector {right arrow over (γ*)}(t) corresponding to the to the modes of the MEG magnetic field of interest {right arrow over (BMEG-OI)}(x, y, z, t). The processor 30 may then derive the MEG signals SMEG from the MEG magnetic field of interest estimates {right arrow over (BMEG-OI-EST)}(x, y, z, t) at the fine magnetometers 26b.
The outside magnetic field {right arrow over (BNON-PHYSICAL)}(x, y, z, t) and MEG magnetic field not of interest {right arrow over (BMEG-NOI)}(x, y, z, t) at the fine magnetometers 26b may simply be ignored, and therefore, not estimated at the fine magnetometers 26b. Alternatively, the processor 30 may estimate the outside magnetic field {right arrow over (BOUT-EST)}(x, y, z, t) at the fine magnetometers 26b by substituting the (x,y,z) locations of the fine magnetometers 26b into the basis functions of the parameterized environmental magnetic field model {right arrow over (BENV-PAR)}(x, y, z, t) corresponding to the modes of the outside magnetic field {right arrow over (BOUT)}(x, y, z, t); i.e., the outside magnetic field estimates {right arrow over (BOUT-EST)}(x, y, z, t) at the fine magnetometers 26b may be recovered from the product of the influence matrix Z and the least squares fit values of the coefficient vector {right arrow over (γ*)}(t) corresponding to the modes of the outside magnetic field {right arrow over (BOUT-EST)}(x, y, z, t). The processor 30 may then use the outside magnetic field estimates {right arrow over (BOUT-EST)}(x, y, z, t) to control the set of magnetic field actuators 28 to at least partially cancel the outside magnetic field {right arrow over (BOUT)}(x, y, z, t), thereby suppressing the total residual magnetic field {right arrow over (BTOT)}(x, y, z, t) to the baseline level at the fine magnetometers 26b.
Referring back to
In particular, the electromagnetic nature of magnetic fields that are generated from electrical sources are different from the electromagnetic nature of magnetic fields that are generated from permanent magnets for a variety of reasons. For example, permanent magnets have a persistent magnetization, and thus, the processor 30 may reduce the content of the outside magnetic field BOUT in the measured total residual magnetic field BTOT-MEAS by eliminating the content of the measured total residual magnetic field BTOT-MEAS corresponding to persistent magnetization. Furthermore, the electrical current along a neural connection that is primarily axial in nature may be distinguishable from a closed electrical current loop, which is more similar to that from a permanent magnet, and thus, the processor 30 may reduce the content of the outside magnetic field BOUT in the measured total residual magnetic field BTOT-MEAS by eliminating the content of the measured total residual magnetic field BTOT-MEAS corresponding to closed electrical current loops. In some cases, there may be closed electrical loops in the brain. However, the scaling of magnetic field delay differs from electrical current in the brain than permanent magnets. Thus, the processor 30 may reduce the content of the outside magnetic field BOUT in the measured total residual magnetic field BTOT-MEAS by eliminating the content of the measured total residual magnetic field BTOT-MEAS corresponding to magnetic fields that decay in space with a scale of that from the permanent magnets.
Thus, referring back to
The processor 30 may be configured for performing the magnetic field distinguishing techniques in any suitable order. Furthermore, the magnetic field distinguishing techniques can be combined as “AND” logic or “OR” logic. For example, is there are conditions A, B, and C that can respectively be associated with the magnetic field distinguishing techniques. Then the processor 30 may accept the portion of the total residual magnetic field BTOT identified as the MEG magnetic field BMEG only if the conditions A-C (or any combination of conditions A-C) are satisfied or may accept the portion of the total residual magnetic field BTOT identified as the MEG magnetic field BMEG if one of the conditions A-C is satisfied. Furthermore, one condition for a magnetic field distinguishing technique may be dynamically varied based on the satisfaction of the satisfaction of the condition of another one of the magnetic field distinguishing techniques. For example, if the condition A for one of the magnetic field distinguishing techniques is satisfied, then the threshold for satisfying the condition B for another one of the magnetic field distinguishing techniques may be lowered.
Thus, it can be appreciated from the foregoing that the signal acquisition unit 18 eliminates large portions of the total residual magnetic field BTOT that do not correspond to the true MEG magnetic field BMEG-TRUE by cleverly combining various signal discriminating techniques, and in particular, based on Maxwell's equations, temporal frequency, spatial frequency, and amplitude.
Referring now to
The method 100 comprises generating the actuated magnetic field BACT that at least partially cancels an outside magnetic field BOUT (e.g., via the set of magnetic field actuators 28 of the signal acquisition unit 18), thereby yielding a total residual magnetic field BTOT (step 102). In the preferred embodiment, the actuated magnetic field BACT is generated in all three dimensions and is uniform, although in alternative embodiments, the actuated magnetic field BACT may be generated in less three dimensions and may be non-uniform (e.g., a gradient).
The method 100 further comprises acquiring the total residual magnetic field measurements BTOT-MEAS respectively at a plurality of detection locations (e.g., from the coarse magnetometers 26a and/or fine magnetometers 26b of the signal acquisition unit 18) (step 104). The method 100 further comprises estimating the total residual magnetic field BTOT-MEAS at at least one the fine detection locations (e.g., at the fine magnetometers 26b of the signal acquisition unit 18) based at least partially on the total residual magnetic field measurements BTOT-MEAS respectively acquired from the detection locations (step 106).
The method 100 further comprises controlling the actuated magnetic field BACT at least partially based on the total residual magnetic field estimates BTOT-EST at the fine detection location(s) in a manner that suppresses the total residual magnetic field BTOT at the fine detection location(s) to a baseline level (by cancelling the outside magnetic field BOUT, e.g., via the coarse feedback control loop 50 and/or fine feedback control loop 52 and sending noise-cancelling control signals C to the set of magnetic field actuators 28 of the signal acquisition unit 18), such that accuracies of the total residual magnetic field measurements BTOT-MEAS acquired at the fine detection location(s) increase (e.g., fine magnetometers 26b of the signal acquisition unit 18 come in-range) (step 108).
In particular, the total residual magnetic field BTOT is suppressed at the fine detection location(s) (e.g., at the fine magnetometers 26b of the signal acquisition unit 18) to the baseline level at the fine detection location(s) by cancelling the outside magnetic field BOUT component relative to the MEG magnetic field BMEG component of the total residual magnetic field measurements BTOT-MEAS acquired from the fine detection location(s) based on a combination of the temporal frequency of the outside magnetic field BOUT (e.g., by suppressing the total residual magnetic field measurements BTOT-MEAS acquired from the fine detection location(s) at DC and harmonic temporal frequencies), the spatial frequency of the outside magnetic field BOUT (e.g., by suppressing the total residual magnetic field measurements BTOT-MEAS acquired from the fine detection location(s) at relatively low spatial frequencies) and/or a strength of the outside magnetic field BOUT (e.g., by suppressing the total residual magnetic field measurements BTOT-MEAS acquired from the fine detection location(s) at relatively high strength frequency components).
Although the outside magnetic field BOUT is at least partially cancelled at the fine detection location(s) by the actuated magnetic field BACT at selected temporal frequencies, spatial frequencies, and/or strengths as a means of suppressing the total residual magnetic field BTOT at the fine detection location(s) to the baseline level, in alternative embodiments, the outside magnetic field BOUT component of the total residual magnetic field measurements BTOT-MEAS acquired from the fine detection location(s) may be suppressed external to the feedback control loop during a post-processing step, in which case, the total residual magnetic field BTOT at the fine detection location(s) may be suppressed to the baseline level utilizing other techniques.
The method further comprises deriving a plurality of MEG signals SMEG respectively from the total residual magnetic field estimates BTOT-EST acquired from the fine detection location(s) (e.g., via the signal acquisition unit 18) (step 110). That is, because the total residual magnetic field BTOT at the fine detection location(s) contains the MEG magnetic field BMEG from the brain 14 of the user 12, and thus by inference, the total residual magnetic field estimates BTOT-EST at the fine detection location(s) contains estimates of the MEG magnetic field BMEG from the brain 14 of the user 12, the MEG signals SMEG can be extracted from the total residual magnetic field estimates BTOT-EST at the fine detection location(s). The method 100 lastly comprises determining the existence and detection location of neural activity in the brain 14 of the user 12 based on the MEG signals SMEG (e.g., via the signal processing unit 20) (step 112).
Referring now to
The method 150 comprises generating a generic model of the environmental magnetic field BENV-MOD in the vicinity of the detection locations, the generic model comprising an initial number of basis functions corresponding to the modes of the environmental magnetic field BENV-MOD (step 152). In one embodiment, the generic model BENV-MOD comprises basis functions for both the outside magnetic field BOUT and the MEG magnetic field BMEG, such that the non-physical portion of the components of both the outside magnetic field BOUT and the MEG magnetic field BMEG can be suppressed in the total residual magnetic field measurements BTOT-MEAS acquired from the detection locations, although in alternative embodiments, the generic model comprises basis functions for only the outside magnetic field BOUT or only the MEG magnetic field BMEG, such that the non-physical portion of the components of either the outside magnetic field BOUT or the MEG magnetic field BMEG can be suppressed in the total residual magnetic field measurements BTOT-MEAS acquired at the detection locations.
The method 150 further comprises applying Maxwell's equations to the environmental magnetic field model BENV-MOD to reduce the initial number of different basis functions, thereby yielding a Maxwell-constrained model of the environmental magnetic field BENV-MAXWELL (step 154). IN the case where the generic environmental magnetic field model BENV-MOD comprises basis functions corresponding to modes of the outside magnetic field BOUT, such basis functions may comprise 0th order basis functions and 1st order basis functions. In another embodiment, the basis functions comprise at least one non-linear basis function (e.g., a vector spherical harmonics (VSH) basis function).
The method 150 further comprises parameterizing the Maxwell-constrained environmental magnetic field model BENV-MAXWELL at least partially based on the total residual magnetic field measurements BTOT-MEAS acquired from the detection locations, and in the illustrated embodiment, based on the total residual magnetic field measurements BTOT-MEAS acquired at the detection locations and the known actuated magnetic field BACT-KNOWN at the detection locations, thereby yielding a parameterized environmental magnetic field model BENV-PAR. In the illustrated embodiment, the Maxwell-constrained environmental magnetic field model BENV-MAXWELL is parameterized by fitting the Maxwell-constrained environmental magnetic field model BENV-MAXWELL to a difference between the total residual magnetic field measurements BTOT-MEAS acquired at the detection locations and the known actuated magnetic field BACT-KNOWN at the detection locations (e.g., using a least squares optimization technique) (step 156).
For example, the coefficients of the basis functions in Maxwell-constrained environmental magnetic field model BENV-MAXWELL may be fitted to the difference between the total residual magnetic field measurements BTOT-MEAS acquired at the detection locations and the known actuated magnetic field BACT-KNOWN at the detection locations, e.g., using a least squares optimization technique. The fitted coefficients may then be incorporated into the Maxwell-constrained environmental magnetic field model BENV-MAXWELL, thereby yielding the parameterized environmental magnetic field model BENV-PAR.
The method 150 lastly comprises estimating the environmental magnetic field BENV-EST at at least one the fine detection locations (e.g., at the fine magnetometers 26b of the signal acquisition unit 18) based on the parameterized environmental magnetic field model BENV-PAR, and in particular, by substituting the fine detection location(s) into the parameterized environmental magnetic field model BENV-PAR (step 158). It should be appreciated that, due to the previous application of Maxwell's equations to the generic environmental magnetic model BENV-MOD, the non-physical portion of the estimated environmental magnetic field model BENV-EST of the measured at the fine detection location(s) is less than the non-physical portion of the environmental magnetic field model BENV component of the total residual magnetic field measurements BTOT-MEAS acquired at the fine detection location(s).
It should be appreciated that, because the non-physical portion of the environmental magnetic field BENV component of the measurements BTOT-MEAS acquired from the location(s) has been reduced by using Maxwell's equations to provide more accurate total residual magnetic field estimates BTOT-EST, the actuated magnetic field BACT, the control of which is at least partially based on the total residual magnetic field estimates BTOT-EST at the fine detection location(s) in the method 100 described above, more accurately cancels the outside magnetic field BOUT at the fine detection location(s), and thus more effectively suppresses the total residual magnetic field BTOT at the fine detection location(s) to the baseline level, such that accuracies of the total residual magnetic field measurements BTOT-MEAS acquired at the fine detection location(s) increase.
Notably, in the case where the Maxwell's equations have been applied to the environmental magnetic field BENV component of the total residual magnetic field measurements BTOT-MEAS acquired from the fine detection location(s) in a manner that reduces the non-physical portion of the MEG magnetic field BMEG component of the total residual magnetic field measurements BTOT-MEAS acquired from the fine detection location(s), the accuracy of the MEG signals SMEG extracted from the total residual magnetic field estimates BTOT-EST at the fine detection location(s) will be increased.
Referring now to
The method 200 comprises generating a generic model of a plurality of magnetic field components of the total residual magnetic field measurements BTOT-MEAS in the vicinity of the detection locations, wherein the generic magnetic field model comprises a plurality of basis functions having multiple sets of basis functions respectively corresponding to modes of the magnetic field components (step 202). In one embodiment, the generic magnetic field model BMOD comprises a coefficient vector and a matrix of influence Z from the coefficient vector to the magnetic field components of the total residual magnetic field BTOT. The coefficient vector has a p number of coefficients respectively corresponding to the basis functions, the influence matrix Z comprises a p number of column vectors and an N number of row vectors respectively corresponding to the total residual magnetic field measurements BTOT-MEAS acquired from the detection locations, and p is less than N.
The method 200 further comprises parameterizing the generic magnetic field model BMOD by simultaneously fitting coefficients of the basis functions of the generic magnetic field model BMOD at least partially to the total residual magnetic field measurements BTOT-MEAS acquired from the detection locations, thereby yielding a parameterized model of the magnetic field components BPAR of the total residual magnetic field BTOT in the vicinity of the detection locations. In the illustrated embodiment, the generic magnetic field model BMOD is parameterized by simultaneously fitting the coefficients of the basis functions at least partially to a difference between the total residual magnetic field measurements BTOT-MEAS acquired at the detection locations and the known actuated magnetic field BACT-KNOWN at the detection locations (e.g., using a least squares optimization technique) (step 204). The coefficients of the plurality of basis functions may be simultaneously fitted at least partially to the total residual magnetic field measurements BTOT-MEAS acquired from the detection locations by equating the product of the coefficient vector and the influence matrix Z to the total residual magnetic field measurements BTOT-MEAS acquired from the detection locations and simultaneously fitting the p number of coefficients in the coefficient vector at least partially to the difference between the total residual magnetic field measurements BTOT-MEAS acquired at the detection locations and the known actuated magnetic field BACT-KNOWN at the detection locations.
The method 200 lastly comprises estimating one or more of the magnetic field components of the total residual magnetic field measurement BTOT-MEAS at each of at least one of the fine detection locations (e.g., from one of the fine magnetometers 26b of the signal acquisition unit 18) respectively based on the multiple sets of basis functions of the parameterized magnetic field model BPAR and in particular, by substituting the fine detection location(s) into the set(s) of basis functions corresponding to the modes of the one or more magnetic field components (step 206). That is, a first one of the magnetic field components of the total residual magnetic field measurement BTOT-MEAS can be estimated at each of at least one of the fine detection locations (e.g., from one of the fine magnetometers 26b of the signal acquisition unit 18) respectively based on a first set of the basis functions of the parameterized magnetic field model BPAR (e.g., by substituting the fine detection locations(s) into the set of basis functions corresponding to the modes of the first magnetic field component); a second one of the magnetic field components of the total residual magnetic field measurement BTOT-MEAS can be estimated at each of at least one of the fine detection locations (e.g., from one of the fine magnetometers 26b of the signal acquisition unit 18) respectively based on a second set of the basis functions of the parameterized magnetic field model BPAR (e.g., by substituting the fine detection locations(s) into the set of basis functions corresponding to the modes of the first magnetic field component); and so on.
In one embodiment, the parameterized magnetic field model BPAR is a parameterized outside magnetic field model BOUT-PAR, the magnetic field components of the total residual magnetic field measurements BTOT-MEAS comprise a physical outside magnetic field BOUT-P component and a non-physical outside magnetic field BOUT-NP of the total residual magnetic field measurements BTOT-MEAS, and the first set of basis functions of the parameterized outside magnetic field model BOUT-PAR corresponds to modes of the outside magnetic field BOUT-P that are physically possible, while the second set of basis functions of the parameterized outside magnetic field model BOUT-PAR corresponds to modes of the outside magnetic field BOUT-NP that are physically impossible. In this case, the physical outside magnetic field BOUT-P component of total residual magnetic field measurements BTOT-MEAS can be estimated at each of the fine detection location(s) based on the first set of basis functions, while ignoring the second set of basis functions, of the parameterized outside magnetic field model BOUT-PAR. The physical outside magnetic field estimates BOUT-P-EST at the fine detection location(s) can then be used in the step 108 of the method 100 as a means to control the actuated magnetic field BACT to at least partially cancel the outside magnetic field BOUT at the fine location(s) in a manner that suppresses the total residual magnetic field BTOT at the fine detection location(s) to the baseline level.
In another embodiment, the parameterized magnetic field model BPAR is a parameterized environmental magnetic field model BENV-PAR, the magnetic field components of the total residual magnetic field measurements BTOT-MEAS comprise the MEG magnetic field BMEG and the outside magnetic field BOUT, and the first set of basis functions of the parameterized environmental magnetic field model BENV-PAR corresponds to modes in the MEG magnetic field BMEG, while the second set of basis functions of the parameterized environmental magnetic field model BENV-PAR corresponds to modes in the outside magnetic field BOUT.
In this case, the MEG magnetic field BMEG component of the total residual magnetic field measurements BTOT-MEAS can be estimated at each of the fine detection location(s) based on the first set of basis functions of the parameterized environmental magnetic field model BENV-PAR, while the outside magnetic field BOUT component of the total residual magnetic field measurements BTOT-MEAS can be estimated at each of the fine detection location(s) based on the second set of basis functions of the parameterized environmental magnetic field model BENV-PAR. The MEG signals SMEG may be derived from the MEG magnetic field BMEG-EST at the detection location(s) external to the feedback control loop in step 110 of the method 100, while the outside magnetic field estimates BOUT-EST may either be ignored or used in the step 108 of the method 100 as a means to control the actuated magnetic field BACT to at least partially cancel the outside magnetic field BOUT at the fine location(s) in a manner that suppresses the total residual magnetic field BTOT at the fine detection location(s) to the baseline level.
In still another embodiment, the parameterized magnetic field model BPAR is a parameterized environmental magnetic field model BENV-PAR, the magnetic field components of the total residual magnetic field measurements BTOT-MEAS comprise a MEG magnetic field of interest BMEG-OI and a MEG magnetic field of not of interest BMEG-NOI, and the first set of basis functions of the generic magnetic field model BMOD corresponds to modes of the MEG magnetic field BMEG-OI of interest, while the second set of basis functions of the generic magnetic field model BMOD corresponds to modes of the MEG magnetic field a BMEG-NOI not of interest. In this case, the MEG magnetic field of interest BMEG-OI component of total residual magnetic field measurement BTOT-MEAS can be estimated at each of the fine detection location(s) based on the first set of basis functions of the parameterized magnetic field model BPAR, while the second set of basis functions may be ignored. The MEG signals SMEG may be derived from the estimates of the MEG magnetic field of interest BMEG-OI at the detection location(s) external to the feedback control loop in step 110 of the method 100.
Although particular embodiments of the present inventions have been shown and described, it will be understood that it is not intended to limit the present inventions to the preferred embodiments, and it will be obvious to those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the present inventions. Thus, the present inventions are intended to cover alternatives, modifications, and equivalents, which may be included within the spirit and scope of the present inventions as defined by the claims.
Pursuant to 35 U.S.C. § 119(e), this application claims the benefit of U.S. Provisional Patent Application 62/975,723, filed Feb. 12, 2020, and U.S. Provisional Patent Application 63/035,683, filed Jun. 5, 2020, which are expressly incorporated herein by reference.
Number | Date | Country | |
---|---|---|---|
62975723 | Feb 2020 | US | |
63035683 | Jun 2020 | US |