Patent Application
20240298950
This document generally relates to magnetocardiography, and more particularly, to estimating maternal and fetal heart signals using magnetic sensors.
In general, the fetal heart rate can be measured by various technologies, which can be categorized into technologies for intermittent and continuous fetal heart rate measurements. They have different clinical objectives: intermittent measurement techniques are used for verification of fetal life or assessment of cardiac performance, whereas continuous measurement techniques allow obstetricians to obtain detailed physiological information about newborns.
Embodiments of the disclosed technology relate to methods, systems, and devices for fetal magnetocardiography (fMCG). The disclosed embodiments advantageously enable long-term monitoring that can be remote from a clinical setting, and are sufficiently portable so that they can operate while the mother engages in substantial motion during daily activity. In an example, this is achieved using magnetic sensors that can operate in a magnetically unshielded environment and do not require cryogenic cooling.
In an example aspect, a system for monitoring a fetal heart signal in a subject includes at least one magnetic sensor configured to measure a combined heart signal comprising a mixture of a maternal heart signal and the fetal heart signal. In an example, the at least one magnetic sensor is configured to operate within a dynamic range from 0.01 pT to 1 mT and with a noise floor within a range from 0.01 pT/√{square root over (Hz)} to 100 pT/√{square root over (Hz)}. The system further includes a processor configured to receive, from the at least one magnetic sensor, the combined heart signal. The processor is then configured to estimate, based on the combined heart signal, one or more parameters associated with a location or a motion of the at least one magnetic sensor, and generate a de-noised combined heart signal by performing a de-noising operation on the combined heart signal. In an example, the de-noising operation comprises a filtering operation configured based on the one or more parameters. Finally, the processor is configured to track, based on the de-noised combined heart signal, an estimate of the fetal heart signal.
In another example aspect, a method for monitoring a fetal heart signal in a subject includes receiving, from at least one magnetic sensor, the combined heart signal, and estimating, based on the combined heart signal, parameters associated with a location or a motion of the at least one magnetic sensor. The method further includes configuring, based on the parameters, a denoising operation, generating a denoised combined heart signal by performing the denoising operation on the combined heart signal, and tracking, based on the de-noised combined heart signal, an estimate of the fetal heart signal.
In yet another example aspect, a method for estimating a fetal heart position includes loading model constraints and initial parameter estimates, localizing sensors relative to beacon locations, generating sensor locations in body co-ordinates, and solving an inverse magnetic model for the fetal heart position.
In yet another example aspect, the above-described method may be implemented by an apparatus or device that includes a processor and/or memory.
In yet another example aspect, this method may be embodied in the form of processor-executable instructions and stored on a computer-readable program medium.
The subject matter described in this patent document can be implemented in specific ways that provide one or more of the following features.
Devices, systems, and methods for determining and monitoring a fetal heart rate signal are described. Section headings are used in the present document to improve readability of the description and do not in any way limit the discussion or the embodiments (and/or implementations) to the respective sections only.
Monitoring fetal heart signals is extremely important for evaluating fetal well-being during pregnancy and labor. Fetal heart rate variability has been shown to be a very important marker of health and is often monitored throughout labor. Measuring the characteristics of individual heart beats is also of clinical value. Tracking fetal movement throughout the day is also a common metric used by doctors to assess health and could be done using heart signals if one could associate them with the heart's orientation. Various modalities have previously been used to monitor fetal heart signals.
During the 19th century, fetal heart rate measurement was performed by stethoscope, which was later improved to an optimized fetal stethoscope. A large improvement in signal quality was achieved in 1960 using fetal scalp electrodes connected during labor, and after the mother's membranes had ruptured. Given the very high SNR achieved, the fetal scalp electrode has remained the gold standard for measuring the comparison performance of other modalities of fetal heart monitoring. However, this method is extremely invasive and can only be used after membranes have been ruptured.
In the 1960s and 1970s, ultrasound Doppler measurements (e.g., as described in U.S. Pat. No. 4,143,650) offered a far less invasive method for identifying and tracking fetal heart rate prior to and during labor, quickly becoming the standard method. While still quite useful in a clinical setting, this method requires significant training to perform correctly and suffers from several drawbacks. For example, it is extremely sensitive to motion artifacts, resulting in a low accuracy of individual R-R pulse intervals (i.e., R-wave peak to R-wave peak, which represents the measurement of the sinus heart period in chronological or heartbeat order). Therefore, data is typically averaged over a 15-second interval. Identifying fetal distress relies significantly upon being able to correctly identify short term deceleration and acceleration of the fetal heart rate, a requirement that is not adequately met by Doppler ultrasound measurements. This makes acquisition take longer and does not represent true real-time tracking. Even under trained clinician control, Doppler ultrasound has a much higher rate of heart rate confusion (identifying maternal heart rate by mistake instead of the fetal heart rate) than more recent modalities.
The next big jump in monitoring capability relied on the use of ECG electrodes to measure potential on the mother's abdomen near the fetus, and on advances in signal processing to separate the stronger maternal heart signal and isolate the fetal heart signal. Multi-electrode systems (e.g., as described in GB 2482758) were first used in clinical settings to monitor mothers shortly before and during labor. These systems provide much better estimates of fetal heart rate, including legitimate single beat timing measurements, such as R-R interval times. The ECG systems also have a much lower confusion rate than the Doppler ultrasound methodology. Further advances in these systems have made them cable-less, allowing mothers to move, shower, and walk while typically using a double belt system to hold the ECG electrodes and communication module. Although these systems represent a big step forward in both measurement performance and comfort for the mother, there are still some difficulties and drawbacks with these systems. In particular, the ECG electrodes must continuously maintain firm contact with the mother's skin. For example, it can be difficult to keep the electrodes adequately attached, especially if the mother moves significantly, there is typically some amount of irritation and redness at the electrode site after a monitoring session, and there are also some distortions in the electrode potential measurements due to the varying dielectric properties of body tissue between the fetal heart and the external electrodes. Furthermore, ECG signals are also significantly attenuated during the vernix period, which occurs after the 27th week of pregnancy. During this period, the ECG signals experience added attenuation from the vernix caseosa, a biofilm covering the fetus which acts as a dielectric.
The ECG modality for identifying and tracking fetal heart rate has recently started moving out of clinical environments and is being pilot tested for remote monitoring under physician care (e.g., as described in U.S. Pat. No. 10,039,459 and US 2021/0259613) in rural areas where it is challenging for mother to access a clinical monitoring site. Other modalities include ultra-wideband (UWB) radar detection (e.g., as described in U.S. Pat. No. 9,078,582) to measure fetal heart rate and its variability.
An alternative to ECG systems are systems that measure fetal heart signals magnetically, termed magnetocardiography (MCG). There are several benefits to MCG over ECG. For example, the magnetic system does not require positive electrode contact with the skin, and magnetic measurements of fetal heart signals are also significantly less distorted by body tissues and are less affected by motion artifacts than ECG measurements.
Fetal MCG systems have previously used SQUID (superconducting quantum interference device) magnetometers (e.g., as described in U.S. Pat. No. 8,374,672) and optically pumped magnetometers (OPMs) in controlled laboratory and specific clinical environments. Both these types of sensors suffer from a very small dynamic range (˜55 nT) which requires them to be operated in a magnetically shielded room. SQUID sensors also require cryogenic cooling with liquid helium or liquid nitrogen while OPMs require precision laser systems to operate.
Therefore, both systems require a footprint which restricts their portability.
The described embodiments benefit from the measurement quality benefits of MCG over ECG, while overcoming the need for restrictive measurement conditions like cryogenic cooling or magnetic shielding. It achieves this by using a different type of magnetic sensor based on effects such as tunneling magnetoresistance (TMR), giant spin resonance (GSR), magneto-impedance (MI) or fluxgate magnetometry. Sensors based on these technologies can measure extremely weak magnetic signals over a much broader dynamic range than SQUID or OPM sensors. Using these high dynamic range sensors advantageously enables additional performance improvements over the ECG modality while enabling the system to be operated in unshielded spaces, including as a wearable device that can operate while the mother engages in significant motion and other everyday activities. The wearable device also represents another step forward in comfort due to the removal of the need for direct skin contact of the sensors, as is required in the ECG modality.
In some embodiments, the systems shown in
Existing systems employ SQUID and OPM sensors, which require magnetically shielded rooms and/or either specialized cooling or electronics systems to operate. Embodiments of the disclosed technology uses magnetic sensors that can operate outside of a magnetically shielded room, since they have a much larger dynamic range, and require no cryogenic cooling. Sensors of this type include those based on tunneling magnetoresistance (TMR), giant spin resonance (GSR), and fluxgate magnetometers.
In some embodiments, a magnetometer used in an example fMCG system has the following properties:
Due to the proximity to both the maternal and fetal hearts, the magnetometer sensors will typically measure some linear combination of the maternal and fetal heart signals, as well as noise. The amount of signal energy from each source will be determined by the sensor's physical proximity to each of the hearts. The described embodiments apply signal processing techniques to separate the maternal and fetal heart rate signals before each of them are processed further.
In some embodiments, the de-noising operation applies bandpass filtering to the time-domain signal. For example, the band between 0.1 Hz-50 Hz may be filtered (using a bandpass filter), and an additional notch filtering operation at 60 Hz (50 Hz in parts of the world) and its harmonics (120 Hz, 180 Hz, etc.) may be applied (using a notch filter) to eliminate noise from the AC wall power. In other embodiments, the bandpass and notch filtering operations may be implemented using wavelet analysis or machine learning techniques, e.g., neural networks.
Motion artifacts. When the mother moves, small changes in the local magnetic field will be observed by the sensors which represent undesirable signals. These changes are termed motion artifacts. The position estimation parameters are used as an input to an adaptive filter which attenuates the motion artifacts, ideally eliminating them.
In some embodiments, the de-noising operation incorporates the position or motion information of the magnetometers and/or subject to filter motion artifacts from the magnetometer signals. In an example, motion information is obtained using an inertial measurement unit (IMU). The IMU obtains data using internal gyroscopes, accelerometers and possibly magnetometers. Motion parameters such as position, velocity and acceleration are estimated from the IMU. The rotational and linear motion parameters can be extracted using techniques such as sensor fusion (e.g., as described in U.S. Pat. No. 10,989,563, which is incorporated, in its entirety, as part of this document). Herein, motion parameters in three-dimensional space may be represented using a quaternion algebra or a combination of Cartesian co-ordinates and Euler angles.
The magnetometer output is fixed to the static reference frame (e.g., as detailed in Section 2.3) using the motion parameters and short-term changes in them. As the mother moves, the magnetic sensors move and rotate as a consequence. The direction in which magnetic signals are observed to point can change, even if the total magnitude and position in space remain constant. These rotational changes are tracked through the motion parameters, allowing the measurements to be related back to the static reference frame established by the system using a simple rotational transform.
In some embodiments, motion artifacts can also be tracked using a “trust mask” which is a metric used by the system to indicate the confidence it has that the magnetic signals observed at some instant in time originate from an actual magnetic source rather than being motion induced. This trust mask can be used for downstream processing, de-weighting data taken during intervals of strong motion.
In some embodiments, source separation techniques are applied to either the output of an individual sensor or a plurality of sensors at once to isolate the maternal and fetal heart signals from the measured superposition signals. In an example, independent component analysis (ICA) or principal component analysis (PCA) may be used to implement source separation. These techniques use a plurality of sensor inputs to separate the signals using a statistical metric. Other methods rely on the input from only a single sensor. For example, neural network (machine learning) approaches, can be used to separate out multiple sources from time-domain signals, rendering them as independent signals.
The human heart rate is variable within the population and the heart rate of mother and fetus are at very different rates. Even for an individual, the heart rate can vary over several minutes due to a variety of normal and abnormal physiological factors. Embodiments of the disclosed technology estimate the heart rate, as well as monitor them over time to track changes therein. Since the maternal heart beats at 60-90 beats per minute (bpm) while the average fetal heart beats at 120-160 bpm, the signals have energy at different frequency bands. Therefore, once the signals have been identified, the described embodiments use a locked loop to identify the fetal and maternal heart beats separately based on their unique cadences.
In some embodiments, the maternal and fetal heartrates are monitored using an initial acquisition and tracking loop model. In an example, the heart rates are obtained sequentially, first tracking the mother or fetus' heart rate before the other. In another example, the two heartrates are tracked in parallel tracking loops or simultaneously within the same loop.
In the acquisition stage, the heart rate is acquired, estimating it from the magnetometer when the heart rate is still unknown. By measuring the magnetic signals over some interval of time, the system makes a statistical estimate of the observed heartrate's parameters, such as its frequency and phase (e.g., what time index corresponds to the R-peak or some other marker). Once this estimate is achieved with a statistical confidence above some threshold, the heart rate is considered acquired. The trust mask obtained from the position data can be used to statically de-weight magnetometer data obtained during periods of motion.
A tracking loop is then initiated after acquisition. Starting with the parameter estimates from the acquisition phase, an adaptive method updates them using incoming heart beats, adjusting them as they change. In some embodiments, a Kalman filter may be employed to update the parameters during the tracking loop.
The heart rate parameters can be used to identify time intervals over which the individual heartbeats are expected to be observed in the magnetic data. These individual beats can be time sliced using this information. The heart rate parameters themselves can also be used by the system to improve monitoring and/or medical diagnosis.
In some embodiments, known, nearly static reference points can be used to allow the system to relate the coordinate system of the device, to the coordinates of the mother's body and possibly to coordinates in a visual feedback system to the user. Any movement of the subject or the fetus can then be detected relative to this static reference frame.
In some embodiments, alternating current (AC) coils, referred to as “beacons”, can be incorporated into the system, either embedded in a garment or the environment around the subject, as reference locations. The coils are controlled in such a manner that their signals can be unambiguously detected by the magnetic sensors. In an example, the coils can be operated simultaneously, whereas in another example, they are operated sequentially. In both examples, the coils can be unambiguously identified with that reference location by the system.
In some embodiments, the coils may be operated at independent frequencies, making the signal from each coil distinguishable from all others. In other embodiments, the coils may emit a spread spectrum signal, with each coil emitting a unique identifier such as a pseudo-random number, like in GPS. Alternatively, the coils may be operated simultaneously at the same frequency, which treats the coils as spatially distributed. In some embodiments, the reference coils are operated during the entire measurement, whereas in other embodiments, they are operated only some of the time.
In some embodiments, the maternal heart signal can be used in much the same manner as a beacon source. It can be assumed to be in a fixed position relative to the fetal heart rate signal and even relative to any shifting of the beacons or sensors. Therefore, it acts as a marker relating the garment co-ordinates to the physical co-ordinates of the mother's body.
In some embodiments, and as illustrated in
Embodiments of the disclosed technology are configured to use a reference frame, which provides a clear and effective way to transmit guidance to the user on how to adjust the position of the sensors, as illustrated in
Embodiments of the disclosed technology are configured to track the position of the fetal heart. If the locations of the sensors are first determined, the position and strength of the fetal heart magnetic source distribution are determined using an inverse method. An example process flow is illustrated in
In an example, a sensor's location relative to a sensor beacon of known strength whose magnetic field is approximated as a magnetic dipole source meaning that the field a position r from the dipole obeys the equation
The location of the sensor {right arrow over (r)} can be found using the inverse of the above equation.
In some embodiments, a total of five measurements are required. The location of the sensor and the dipole strength vector can be described by the parameter set (x, y, z, θ, ϕ, m) where x, y, and z describe the location of the sensor relative to the dipole in Cartesian co-ordinates, θ and ϕ describe the angular orientation of the dipole relative to the sensor and m is the dipole strength. As the dipole strength m is already known, there are five remaining unknown parameters to be solved. In embodiments where a three-axis magnetic sensor is used, the five measurements require the use of two sensors, since three measurements, in x, y, and z, are taken by the sensor for each beacon.
In some embodiments, an iterative method can be used to solve the inverse problem, as illustrated in
Herein, the gradient of the error is used to update the parameter values. The derivative with respect to each parameter is calculated. The parameters are updated by the negative gradient times some proportionality constant (known as the learning rate). Some embodiments can separate the search into a coarse and fine grain method. The difference between the two steps being the magnitude of the learning during each iteration. The coarse stage will use a larger learning rate than the fine stage. If the error is below a coarse grain threshold, the coarse update is skipped. Iteration at the fine update size continues until a second, smaller fine grain threshold is reached at which point, the iterations stop.
The heart can be approximated as a magnetic dipole source following the formula given in Section 2.5.1. Using the readings from multiple sensors, the inverse of this equation can be solved to find the location of the dipole relative to the sensors and any beacons in the system.
In some embodiments, at least two measurements along each of the x, y, and z axes are required, for a total of six measurements. With at least six measurements, the location of the dipole in 3D space, the dipole strength, and the orientation of the dipole can be computed. Here, the orientation refers to the θ and ϕ angles in a spherical co-ordinate system. Each dipole is then described by the parameter set (x, y, z, θ, ϕ, m) where m is the magnitude of the dipole.
In other embodiments, only five measurements are required by not measuring the θ angle, since a dipole source's field is symmetric in θ. Here, the relative orientation of the dipole to the sensors can still be measured. However, ignoring the θ coordinate comes at the cost of being able to measure the relative distance to other dipole sources, e.g., the maternal heart signal.
In some embodiments, the dipole location can be solved by an iterative method, as illustrated in
Herein, the gradient of the error is used to update the parameter values. The derivative with respect to each parameter is calculated. The parameters are updated by the negative gradient times some proportionality constant (known as the learning rate). Some embodiments can separate the search into a coarse and fine grain method. The difference between the two steps being the magnitude of the learning during each iteration. The coarse stage will use a larger learning rate than the fine stage. If the error is below a coarse grain threshold, the coarse update is skipped. Iteration at the fine update size continues until a second, smaller fine grain threshold is reached at which point, the iterations stop.
The method 1000 includes, at operation 1020, estimating, based on the combined heart signal, parameters associated with a location or a motion of the at least one magnetic sensor, and at operation 1030, the method 1000 includes configuring, based on the parameters, a de-noising operation. Determining location and/or motion parameters and performing de-noising is detailed in Section 2.2.1.
The method 1000 includes, at operation 1040, generating a denoised combined heart signal by performing the denoising operation on the combined heart signal.
The method 1000 includes, at operation 1050, tracking, based on the de-noised combined heart signal, an estimate of the fetal heart signal.
In some embodiments, the fetal heart rate signal can be isolated by performing a source separation operation on the denoised combined heart signal to generate an estimate of the maternal heart signal and an estimate of the fetal heart signal. The source separation operation, which may be implemented using ICA or PCA on signals from single or multiple sensors, is detailed in Section 2.2.3. Alternatively, the fetal heart rate signal is tracked in the denoised combined heart signal using a tracking loop and without explicitly separating the maternal and fetal heart rate signals.
The described embodiments provide the following systems, methods, and devices directed to fetal magnetocardiography:
In this patent document, the described technical solutions are configured to monitor the maternal and/or fetal heart rate. In some examples, monitoring the heart rate includes continually (or near continually) collecting data points associated with the maternal heart rate and/or fetal heart rate. In some examples, monitoring the heart rate includes tracking the heart rate, which may include both continually (or near continually) collecting data points associated with the heart rate and using a tracking loop, e.g., separate tracking loops to track the maternal heart rate and the fetal heart rate.
The following solutions for fetal magnetocardiography, which may be combined with the aspects enumerated above, may be preferably implemented by some embodiments.
Implementations of the subject matter and the functional operations described in this patent document can be implemented in various systems, digital electronic circuitry, or in computer software, firmware, or hardware, including the structures disclosed in this specification and their structural equivalents, or in combinations of one or more of them. Implementations of the subject matter described in this specification can be implemented as one or more computer program products, i.e., one or more modules of computer program instructions encoded on a tangible and non-transitory computer readable medium for execution by, or to control the operation of, data processing apparatus. The computer readable medium can be a machine-readable storage device, a machine-readable storage substrate, a memory device, a composition of matter effecting a machine-readable propagated signal, or a combination of one or more of them. The term “data processing unit” or “data processing apparatus” encompasses all apparatus, devices, and machines for processing data, including by way of example a programmable processor, a computer, or multiple processors or computers. The apparatus can include, in addition to hardware, code that creates an execution environment for the computer program in question, e.g., code that constitutes processor firmware, a protocol stack, a database management system, an operating system, or a combination of one or more of them.
A computer program (also known as a program, software, software application, script, or code) can be written in any form of programming language, including compiled or interpreted languages, and it can be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment. A computer program does not necessarily correspond to a file in a file system. A program can be stored in a portion of a file that holds other programs or data (e.g., one or more scripts stored in a markup language document), in a single file dedicated to the program in question, or in multiple coordinated files (e.g., files that store one or more modules, sub programs, or portions of code). A computer program can be deployed to be executed on one computer or on multiple computers that are located at one site or distributed across multiple sites and interconnected by a communication network.
The processes and logic flows described in this specification can be performed by one or more programmable processors executing one or more computer programs to perform functions by operating on input data and generating output. The processes and logic flows can also be performed by, and devices can be implemented as, special purpose logic circuitry, e.g., an FPGA (field programmable gate array) or an ASIC (application specific integrated circuit).
Processors suitable for the execution of a computer program include, by way of example, both general and special purpose microprocessors, and any one or more processors of any kind of digital computer. Generally, a processor will receive instructions and data from a read-only memory or a random-access memory or both. The essential elements of a computer are a processor for performing instructions and one or more memory devices for storing instructions and data. Generally, a computer will also include, or be operatively coupled to receive data from or transfer data to, or both, one or more mass storage devices for storing data, e.g., magnetic, magneto-optical disks, or optical disks. However, these are optional. Computer readable media suitable for storing computer program instructions and data include all forms of nonvolatile memory, media and memory devices, including by way of example semiconductor memory devices, e.g., EPROM, EEPROM, and flash memory devices. The processor and the memory can be supplemented by, or incorporated in, special purpose logic circuitry.
While this patent document contains many specifics, these should not be construed as limitations on the scope of any invention or of what may be claimed, but rather as descriptions of features that may be specific to particular embodiments of particular inventions. Certain features that are described in this patent document in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable subcombination. Moreover, although features may be described above as acting in certain combinations and even initially claimed as such, one or more features from a claimed combination can in some cases be excised from the combination, and the claimed combination may be directed to a subcombination or variation of a subcombination.
Similarly, while operations are depicted in the drawings in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed, to achieve desirable results. Moreover, the separation of various system components in the embodiments described in this patent document should not be understood as requiring such separation in all embodiments.
Only a few implementations and examples are described, and other implementations, enhancements, and variations can be made based on what is described and illustrated herein.
This patent document claims priority to and benefits of U.S. Provisional Patent Application No. 63/489,039, filed on Mar. 8, 2023. The entire content of the before-mentioned patent application is incorporated by reference as part of the disclosure of this patent document.
| Number | Date | Country | |
|---|---|---|---|
| 63489039 | Mar 2023 | US |
