Patent Application
20090012387
The invention relates to the handling of electric and electromagnetic signals inside a room with a nuclear magnetic resonance (MR) apparatus such as a scanner or a spectrometer. In particular, the invention relates to recording of physiological signals from a patient during scanning.
Sometimes, it is necessary to acquire additional data during magnetic resonance scanning or spectroscopy, typically in the form of non-MR electric or electromagnetic (EM) signals. These data range from electrophysiological (EPH) signals such as physiological time courses (e.g. electroencephalograms (EEG), electrocardiograms (ECG), blood pressure, respiration etc) to subject responses (e.g. keystrokes, joystick movements). As will be described below, several issues complicate the acquisition of such additional data in the environment of a MR apparatus.
There exist examples of acquisition of radio frequency (RF) signals induced by the field of the MR apparatus that indicate values of local parameters in the subject or object. Here, a quartz crystal is used as a sensor for the local RF field strength (Wang and Leigh, Magn. Reson. Med. 33 (1995), p. 843) or for the local temperature (Wang and Leigh, 3 magn. reson. B, 105 (1994), p. 25 and Simon, J. magn. reson. 128 (1997), p. 194). Both applications make use of the excitation of resonance frequencies in quartz crystals by the RF excitation pulses from a MR apparatus. The excited resonance signals can subsequently be detected by the coils of the MR apparatus. The amplitude of the resonance signal is proportional to the field strength of the RF excitation pulses and provides a measure of the field strength from the MR apparatus in the quartz crystal. The frequency of the quartz resonances varies with the temperature the crystal and the frequency of the resonance signals thus provides a measure of the temperature of the quartz crystal.
In the field and temperature sensors described in the above, the quartz resonance signal is not used to convey data in the form of signals or messages; the signal only reflects the local environment inside the crystal. To convey additional data such as a data sequence or a message, this does not suffice.
Several issues complicate the acquisition of this additional data in the faraday cage of an MR spectrometer or an MR imaging (MRI) suite. First, the additional non-MR apparatus in the scanner room may create unwanted distortions in the MR-images, known as image artifacts. As a result, the sampling equipment needs to be MR compatible, meaning safe, well-functioning at high field, insensitive to radio frequency waves and electromagnetically silent at MR frequencies. Getting the acquired additional signal out of the scanner room is a second problem. This either requires fibre optics or filters that can let the signal out while keeping the Faraday cage intact. A third problem has to do with the MR-scanner disturbing the acquired signal by generating EM noise in the signals, referred to as signal artifacts. Recording of electrophysiological signals during MRI poses extra difficulty as the rapidly changing magnetic fields induce electrode voltages that are often orders of magnitude stronger than the weak electrical signals originating from neural activity. This noise is in particular generated during the fast changes in magnetic gradients needed for MR-scanning. Traditionally EPH-recordings have therefore been performed in silent periods during the scan, where images were not acquired. However, functional MRI (fMRI) mostly used to measure brain function, for example, benefits from imaging with high temporal resolution, and unnecessary pauses should normally be avoided.
Recording of EPH signals and subject responses may for example represent responses to stimuli presented to a patient in the scanner. They often need to be correlated with MR imaging or spectroscopy, thus adding the complexity of synchronization between MR data and non-MR data. When the data are to be analysed, a fifth problem arises. The MR and additional data are stored in different equipment (e.g. on one computer per extra signal type), making data storage, recalling and correlation troublesome.
A particularly interesting and demanding example is recording of electroencephalography (EEG) during fMRI, see e.g. U.S. Pat. No. 5,445,162 and Ives et al., Electroencephalography and Clinical Neurophysiology, 87, 417 (1993). This combination has great potential ranging from patient supervision, over research in basic neurological processes to improved diagnosis of epilepsy. However, EEG recording in an MRI environment suffers extreme degradation due to pulse and imaging artifacts that are often orders of magnitude larger than the EEG signal of interest. The imaging artifacts are induced by the strong and rapidly changing gradients used for fast imaging. The corresponding artifacts can be reduced somewhat by methods limiting the presence of current loops, e.g. by use of special montages and a common reference or dual-lead electrodes, lead-twisting and stable support of the electrodes. In U.S. Pat. No. 5,445,162 and Ives et al., a careful selection and arrangement of analogue multiplexed cable-telemetry equipment has been used to eliminate ferrous materials and RF sources in the EEG equipment.
Filtering techniques, e.g, moving average template subtraction followed by adaptive noise reduction, have also been used to reduce the pulse, motion and imaging artifacts to acceptable levels, but there is a limit to the precision determined by the shot-to-shot variation in the distortion. Furthermore, removal of large noise components from small signals is demanding in terms of bandwidth, timing control, digital resolution and linearity. As distortions and their variation increase with field/gradient strength, imaging speed and patient motion, moving average filtering alone does not provide satisfactory EEGs in all situations.
These problems also pertain to the EEG recording method disclosed by Cohen in US 2004/0097802 involving the synchronization of high-performance electrophysiological equipment with the MR scanner using a clock signal from the scanner. Using the achieved, very precise synchronization, the repetitive gradient noise from the scanner can be estimated and subtracted.
An alternative “stepping stone sampling” technique of particular interest in the present context was introduced by Anami et al (Neuroimage 19, 281), who demonstrated that EEG can be recorded with limited distortion in the short periods between gradient reversals during echo-planar imaging (EPI). The method benefits from synchronization between scanner and EEG clocks. Triggered 20 kHz sampling was used to record the EEG during 400 μs silent periods where also the MR signals were measured, i.e., in the periods between changes of directions of k-space-traversal in blipped EPI. This method is highly suited for fMRI.
The setup used by Anami et al. is fairly complicated, and relies on microsecond synchronization between scanner and EEG equipment obtained by driving the EEG-system with the scanner clock and using high bandwidth sampling and triggered sample-hold. Also, the MR sequences generally have to be edited to provide the needed triggering. After acquisition, the problem remains of correlating EEG and MR images stored and time-stamped differently in separate computer systems.
It is an object of the invention to provide a method and a system for recording electric and electromagnetic non-MR data signals using a MR apparatus.
It is another object of the invention to provide a method and a system for synchronising the recording of MR data signals and non-MR data signals in the MR apparatus.
It is yet another object of the invention to provide a method and a system for easing the recording of electrophysiological (EPH) signals from a scanned subject during scanning.
The problems related to artifacts in MR data signals and non-MR data signals are present in both MR scanners and NMR spectrometers, although they relate to very different fields. In the present description, the term MR apparatus covers both MR scanners used for imaging, NMR spectrometers used for spectroscopy and any other application or apparatus for exciting and detecting precessing nuclear spins. In the present specification, the term MR apparatus designates an apparatus having several parts, some typically situated inside the faraday cage, some typically situated outside. The MR apparatus includes at least the following parts:
Also, in the present context, a MR signal or MR data signal is any signal which is generated by precessing nuclear spins and recorded by a receiving coil as well as signals derived from the recorded signal, e.g. by sampling, modulation, multiplexing, transformation, conversion, etc. The signal path for a MR signal in the MR apparatus starts at the receiving coils of the MR apparatus and ultimately ends at the storage medium holding the stored MR data. The signal carrying the MR data may thus be manipulated, filtered, converted etc. several times along the MR signal path.
The present invention attains the objects mentioned above and other objects by exploiting MR apparatuses capabilities of handling signals in the radio frequency (RF) range. As MR signals from the precessing nuclear spins are relatively weak RF signals, MR apparatus are equipped with high quality components for fast and extremely precise sampling and recording of RF signals.
By intentionally modulating other signals holding non-MR data onto a signal having a frequency in the range of one or more channels of the MR apparatus MR apparatus, the resulting signal can be controllably received by the MR apparatus. If this signal is an RF signal received by the receiving coils of the MR apparatus, it will generate a large but controlled (image) artifact in the recorded MR data. By selecting a proper frequency range and modulation scheme for the modulation onto the RF signal, the artifact from these signals will not interfere with the recorded MR data. The non-MR data signals may e.g. be modulated onto an RF carrier with a frequency corresponding to the edge of an image or to a position in a spectrum that is not typically used for identification. The resulting artifact is thereby recorded and saved in the MR apparatus and may be restored at any time. Preferably, the non-MR data signal is amplitude modulated onto a carrier signal, however, frequency- or phase modulation may be applied as well as multiplexing of several non-MR data signals on the same carrier. Also, modulation is the use of a data signal to actively and deliberately modify or adjust another signal referred to as a carrier (typically of higher frequency) so that the data signal can be restored later by demodulation.
In the present context, non-MR data or non-MR data signal is any signal which is not directly generated or induced by the MR apparatus. The non-MR data signals are signals from another apparatus or devices separate from the MR apparatus. In order to be recorded by the MR scanner as described in the above, the non-MR data signals are modulated onto a carrier signal in the RF band of the MR apparatus, and the modulated carrier signal is then transmitted to and recorded by the MR apparatus. The signals modulated onto the RF carrier may e.g. be EPH signals such as EEG or ECG signals, signals resulting from monitoring of blood pressure, respiration etc., or signals representing subject responses such as keystrokes, joystick movements or speech.
Thermal, optical and other non-electrical signal do not suffer from the same limitations inside a MR suite as electrical signals do. For this reason, the non-MR data signal is preferably an electric signal, such as a signal resulting from a measurement of a value or the detection of an event using an electrical apparatus separate from the MR apparatus.
Signals from precessing nuclear spins in the scanned subject or object, which are induced directly by the magnetic and electromagnetic fields from the MR apparatus are not non-MR data signals according to the above definition. However, these signals may be used as the carrier signals onto which the non-MR data signal may be modulated. In this case, an apparatus or a device separate from the MR apparatus may use the non-MR data signal to modulate the magnetic or RF fields in a local region in the MR receiver coil so that the non-MR data is encoded in RF waves emitted from the region at the Larmor frequency.
Hence, there are three types of signals to distinguish between in the present context;
Unless a distinction is required by the circumstances, both the non-MR data signals (type 1) and the non-MR data modulated RF carrier signals (type 2) will be collectively referred to as non-MR data signals as the first is often contained in the second by modulation. Also, when it is clear from the context, these signals will also be referred to simply as electric or EM signals.
It is an advantage of the invention that the recording of non-MR data signals during a MR sequence does not lead to uncontrollable artifacts on recorded MR data signals and vice versa. As the non-MR data are recorded without any dedicated equipment, the invention minimises the need for costly special equipment adapted for use in MR environments. Also, no new signal lines are needed for transferring data out of the faraday cage or scanner room.
Since the MR apparatus performs the recording of the non-MR data signals, the recording of non-MR data signals is inherently synchronized with the MR measuring sequence. This provides the advantage of reducing artifacts on the non-MR data signals created by the changes in magnetic gradients from the MR scanner. Also it facilitates the subtraction of residual gradient artifacts.
It is an added bonus that the non-MR data corresponding to a MR measuring sequence is saved together and in synchronisation with the MR data from the sequence, which simplifies the task of keeping track of files and synchronisation.
According to a first aspect, the present invention provides a method for generating, transmitting and recording a non-MR data signal during a magnetic resonance sequence, the method comprising the steps of:
Preferably, the step of modulating the non-MR data comprises modulating the non-MR data to an electric or EM carrier signal having a predetermined frequency in the RF band. The carrier signal is typically provided by a clock signal generator or an oscillator in a dedicated modulator. The frequency of the carrier signal is important since it differentiates the non-MR data signals from the MR data signals and thereby allows restoration of the non-MR data even though the signals have been handled together by the MR apparatus. Hence, the predetermined frequency is preferably distinguishable from the frequencies of MR signals in the sequence, e.g. by lying outside or on an unused interval of the RF band used by the MR signals. The specific frequencies and the required distance from frequencies occupied by MR signal depends on a number of factors such as the type of pulse sequence run by the MR apparatus, the object in the scanner, the type of non-MR data, the modulation scheme of the non-MR data to the carrier signal, etc.
To ensure that the MR apparatus will transmit the received non-MR data signal undistorted, the carrier signal preferably has a frequency in the range of one or more channels of the MR apparatus.
The non-MR data modulated RF signal may be transmitted in a wire or cable connected to a receiving part of the MR apparatus, or wirelessly to a receiving coil of the MR apparatus. In any case, the electric or EM, Non-MR data modulated RF signal is preferably introduced into the signal path for MR data signals at a point residing inside the magnetic resonance apparatus room.
During periods of a MR recording pulse sequence, the variations in the magnetic gradients in the scanner are so strong that many electrical signals will be affected. At other periods, the magnetic gradients are kept at least substantially constant, and here, handling of other electrical signals in or near the scanner is possible. It can therefore be important to limit the bandwidth of the non-MR data by gating to remove noise in the non-MR signal by band-pass filtering (e.g. thermal). Hence, the method may further comprise the step of gating said electric or EM non-MR data signals in predetermined periods of the MR sequence (sample-hold). Alternatively or supplementary thereto, as the sequence typically applies those ‘silent’ periods for recording MR-data signals, the method may instead comprise the step of synchronising a measuring of said electric or EM non-MR data signals with a measuring of a MR data signal obtained during the MR sequence. Also, after digitalization, the non-MR data is preferably recorded in synchrony with MR data obtained during the same MR sequence. This allows correlation between e.g. the activity determined from an EEG and the corresponding fMRI data.
Typically, the RF modulated non-MR data is received and recorded at the MR apparatus during a MR sequence and together with a MR data signal from a scanned subject or object. Even if the recording sequences of MR and non-MR data does not overlap in time, the invention still presents a huge advantage since the MR apparatus replaces some of the additional apparatuses involved with the non-MR data. In the prior art, each type of non-MR data signal would involve at least a sampler, filters (for getting signals out of the cage). These apparatuses would have to be designed for use inside a MR room or turned off and possibly removed from the room during scanning. According to the present invention, the MR-apparatus samples the non-MR data signals and transmits out of the cage through high quality filters. A lot of the equipment specially designed for MR environments are therefore no longer necessary with the principles of the present invention.
In the field of MR imaging, the scanned subject is often a human or animal in which case it is often relevant to monitor the physiological state of the subject during scanning. Thus, according to a second aspect, the invention provides a method for obtaining a physiological signal of a subject positioned in a magnetic resonance scanning room during a magnetic resonance scanning sequence, the method comprising the steps of:
The sensed and/or monitored physiological signal may be an electric or electromagnetic physiological signal, or the sensor may be an electric or electromagnetic sensor so that the output physiological signal is an electric or electromagnetic physiological signal. Alternatively, the output physiological signal may be in the form of an optical signal, such as a light signal or a plasmonic signal, in which case the method further comprises the step of converting the sensed and/or monitored physiological signal to an electric or electromagnetic physiological signal. The conversion may be performed before or simultaneous to the RF modulation using an appropriate optoelectronic modulator.
The method may further comprise the steps of amplifying the physiological signal before the RF modulation with an appropriate electric or optic amplifier and transmitting the sensed and/or monitored physiological signal to the RF modulator.
In a third aspect, the invention provides a MR apparatus for implementing the above methods, i.e. a MR apparatus for receiving electric or EM carrier signals modulated with non-MR data during a MR sequence and restoring the non-MR data, the MR apparatus being configured to receive non-MR data modulated carrier signals and transmit the received signals along a signal path of MR signals, the MR apparatus comprising software for accessing received and digitised MR signals and non-MR data modulated carrier signals and, using knowledge of a carrier frequency of the non-MR data modulated signals, identifying non-MR data modulated carrier signals and, using knowledge of the modulation of said non-MR data modulated signals, demodulating identified non-MR data modulated carrier signals and restoring the non-MR data or a function thereof.
Preferably, the MR apparatus further comprises software for synchronising a time stamp on the restored non-MR data with a time stamp on corresponding MR data.
In a fourth aspect, the invention provides the use of a prior art MR apparatus for implementing the above method, i.e. the use of a MR apparatus for receiving, transmitting, digitizing, identifying and demodulating non-MR data modulated carrier signals generated inside a magnetic resonance room during a MR sequence.
Preferably, the MR apparatus is also used to measure the non-MR data modulated signals in synchronisation with predetermined periods of the MR sequence or with measuring of MR data by the MR apparatus.
As the RF signals generated by precessing nuclear spins are very weak, MR apparatuses are very sensitive. All MR apparatus will therefore record almost any non-MR RF signal present as noise (which is of course why the scanner is situated in a Faraday cage). According to the present invention, it is the high sensitivity towards RF signals which is used when non-MR RF signals are controllably generated, and recorded by the MR apparatus.
The methods according to the present invention are novel and inventive over recording of RF noise in the prior art in that the non-MR data signals are deliberately and intentionally modulated to a RF frequency which is within the range of the MR apparatus but outside or in the outskirts of the frequencies of the MR signals of relevance. The non-MR RF signals according to the invention are therefore not equivalent to RF noise or other unwanted signals since they appear in well defined regions in an image or spectrum, and since they are prepared for demodulation for restoration of the non-MR data.
Although other MR apparatuses have been used to record RF noise, the apparatuses according to the present invent are novel and inventive over corresponding apparatuses of the prior art in that it is modified, prepared, adapted or suited for, or comprise means for, or is used in identifying non-MR signals introduced in the MR data signal path, and demodulating the non-MR data signals to restore the non-MR data.
Typically, the present invention will be carried out using a system which is a combination of a modulator for modulating the non-MR data onto signals which can be handled by the MR apparatus, a MR apparatus, and software for performing the identification and demodulation of the non-MR data signals. The system may consist of separate parts or may be build-in and integrated in a MR apparatus. According to a fifth aspect, the invention provides a system for recording non-MR data generated inside a MR measurement room, the system comprising a modulator to be positioned inside the MR measurement room and a MR apparatus, the modulator comprising
The software may be associated with the MR apparatus in that it is to be executed by a computer forming part of the MR apparatus, or in that it is to be executed by a computer to which at least the digitised non-MR data can be transferred from the MR apparatus.
In a preferred embodiment, the modulator is partly integrated with the MR apparatus in that it is integrated with a detection coil for a MR apparatus. Therefore, in a sixth aspect, the invention provides a coil for a MR apparatus, the coil being configured to receive and introduce electromagnetic, radio frequency (RF) MR signals and electric non-MR data signals into a signal path of a MR apparatus, the coil comprising
a receiving coil part for receiving electromagnetic RF MR signals, generating corresponding electric RF MR signals,
a socket part for receiving electric non-MR data signals,
a modulator electronically connected to the socket part, the modulator comprising
Modulators for modulating an incoming signal onto an RF carrier signal exist in the prior art. The application of such a modulator to modulate incoming data to a RF signal which is transmitted to and recorded by a MR apparatus has not been disclosed in the prior art. Hence, according to a seventh aspect, the invention provides the use of a modulator for receiving and modulating non-magnetic resonance (non-MR) data generated inside the scanner room onto a magnetic, electric or electromagnetic signal to form a modulated carrier signal and transmitting the modulated carrier signal to a magnetic resonance apparatus, the modulator being positioned inside a magnetic resonance apparatus room and operating during a magnetic resonance sequence, the modulator comprising
The present invention provides a new and inventive solution to the problems of image and signal artifacts, synchronization, data management and bandwidth presented in relation to the prior art. The invention provides a particularly simple method for recording non-MR data signals. In a preferred embodiment, the non-MR data or data signals is modulated onto RF signals which can be received by the MR apparatus together with the normal MR signals. Here, the RF carriers are emitted from a frequency generator and are modulated by electrical signals that are to be measured during imaging. In an alternative embodiment, the modulated carrier signals are fed to the MR apparatus at a later stage in the signal path of the MR apparatus.
These modulated RF signals emitted within the RF enclosure are recorded by the MR apparatus and can subsequently be extracted from the MR images or spectra. The modulated signals can be emitted as radio waves inside the RF enclosure by use of a simple aerial, thereby ensuring galvanic separation between the subject and the recording apparatus. Exploiting surplus bandwidth, the MR system is therefore used for recording and storing both MRI and electrical signals with inherent microsecond synchronization.
In order to limit the bandwidth of the non-MR signal and/or avoid the gradient and RF induced noise from causing transients in the measured EPH-signals, gradient activity triggered gating can be implemented with a sensor coil placed near the opening of the scanner.
In addition to the MR equipment, only relatively simple, light, safe and low-cost equipment is needed. If raw data is available from the scanner, sequence modification is normally not needed. The method is highly generic and is compatible with most existing MR systems with no need for hardware adaptation or interfacing.
An added, important advantage is that the technique offers an implementation of EEG-recording techniques presented by Anami et al (Neuroimage 19, 281) and Cohen et al. (Proc. of Human Brain Mapping Annual Meeting. 2001; 6). It is therefore considered highly relevant for EEG-fMRI.
In the instrument visualization described previously, self-resonant RF circuits are attached to instruments to act as high-contrast markers for localisation purposes in MR imaging (MRI), see e.g. Weiss et al., Proc. 9th Annual Meeting on ISMRM, 2001, p. 544 or Wong et al., J magn. reson. Imaging 12 (2000), p. 632. Here, a RF coil is tuned to the Larmor frequency of the scanner modifies the effective excitation angle for the substance inside the coil, whereby the interior of the coils can be brightly depicted. In order to turn the visualisation on/off, the RF circuit can be tuned/detuned by an optical signal from the scanner using a photodiode connected to the RF circuit. Turning something on/off is only a modulation if it forms part of an encoding scheme where a series of on/off's are provided to convey additional data (e.g. in the form of binary data). This is clearly not the object in the instrument visualization applications. Hence, no data is modulated onto the MR signal from the interior of the coil and as a consequence, there is no demodulation of data or data signals from the acquired image.
The applications of quartz crystals as temperature sensors described previously make use of inherent-material responses to the fields from the MR apparatus. It does not present a deliberate modulation of a signal onto a carrier, instead, the carrier is consequently and inevitably formed with this variation. This is the same way as the resonance frequencies and relaxation times of nuclear species depends on its molecular environment, e.g. oxygenated or deoxygenated hemoglobin.
In the following, the invention and selected embodiments thereof will be described in relation to the accompanying figures in which:
In a preferred embodiment, the present invention provides a method and a system for recording EPH signals, subject responses and other signals to be generated inside a MR room and recorded during a MR scanning or spectroscopy. In other aspects, the invention relates to an MR apparatus or a receiving coil of an MR apparatus configured to perform such recording. In yet other aspects, the invention relates to the use of a MR apparatus or a modulator for making such recordings. These aspects will all be explained in the following in relation to a number of preferred embodiments.
In a first preferred embodiment, the method for recording non-MR signals according to the invention can involve a number of distinct steps described in the following with reference to
Steps I and II take place in a pre-processing unit 1 for detecting and processing non-MR signals.
Analogue or digital non-MR signals are generated by an output from a signal source. Examples:
The non-MR signals may be prepared for further processing, e.g., by any combination of amplification, conversion, splitting, combining, digitization, mixing, gating, delaying, filtering etc. The process possibly involves a conversion of the signal to an electric signal. Examples:
Step III takes place in a modulator 2, the non-MR data signals are encoded onto one or more electric or electromagnetic carrier signals with a frequency which can be handled by the relevant part of a MR apparatus 4. The one or more modulated electric or EM RF carrier signals are transmitted to the MR apparatus by a wired connection (electrical) or wirelessly (electromagnetic). Examples:
The non-MR data signals are introduced somewhere along the normal scanner/spectrometer signal line, and are possibly digitized. They may, for example, be measured during scanning using a normal MR spectroscopy or imaging sequence and receiver coil, so that the non-MR signals are embedded in the (complex) MR raw data or images. Alternatively, if a relatively low frequency (e.g., 100 kHz) is chosen in step III, the signal can be mixed into the MR signal path at a low frequency stage immediately prior to digitization.
As the non-MR signals are introduced in the MR data signal path, they are processed similarly to the MR data signals and thereby possibly embedded in the MR raw data or images.
The original non-MR signals generated in Step I, or derived functions of these, are extracted from the MR raw data after digitization. Scaled and offset sampled intervals of the original non-MR signals may, for example, be derived from imaging raw data.
The method for recording non-MR signals outlined above can be implemented by existing MR apparatuses or by new MR apparatus which have been designed to implement this method.
The images and electrical signals are stored, e.g. together in a radiological picture archiving system (PACS) where the scanner may send the images anyway. The DICOM data format (Digital Imaging and Communications in Medicine) used by most PACS implementations supports both EPH- and MR-image formats.
In a first embodiment, the above method is implemented by existing MR apparatus, an amplifier with few modifications as well as one specially designed modulator. In the following, the equipment applied in the first embodiment is described in detail.
The encoding of the non-MR signal outlined in Step III above can be performed using a specially designed Multi frequency amplitude modulator (MFAM) 2 in
The reference oscillator 21 is a very stable and programmable oscillator made with a voltage controlled crystal oscillator (VCXO) and a low jitter programmable phase locked loop (PLL). This gives the advantage of easy adjustment to the MR frequency range. The clock frequency for each oscillator 22 is obtained by individual low jitter programmable PLLs. The digital output from each PLL is bandpass filtered to avoid problems from higher harmonics. The frequencies of operation are chosen within the bandwidth of the MR apparatus and its components i.e. filters, but outside the frequency range of important MR data signals.
The non-MR input signals of Step I and II are received at inputs 26 and are low-pass filtered by filter 23 to avoid channel interference and distortion of the MR signals. The channel clocks and the filtered signals are fed to amplitude modulators 24. The outputs from all the amplitude modulators are joined in a power combiner 25 and the RF output is sent to the MR apparatus either cable 3 or transmitting aerial 5.
Normally, the measured RF signal from the precessing protons is detected in one or more RF coils 12 of the scanner. The signals from separate coil elements are either combined and processed together, or are processed separately in more or less independent RF channels that may operate at different frequencies. The MR data signals are amplified and modulated down to a relatively low frequency suitable for digitization. This amplification and modulation happens in one or more steps along part 13 of the signal path which may reside both inside and outside of the faraday cage 11. At the boundary of the faraday cage 11, a penetration panel 14 prevents irrelevant RF signals from the outside to be transferred to the faraday cage and detected by the RF coils. As no new signal paths are introduced to transfer the non-MR signal of interest out of the faraday cage 11, the invention involves no alterations to the penetration panel. Before storage in memory 16, all signals in the signal path, i.e. both MR and non-MR if present, are digitised in analogue to digital converter (ADC) 15. Eventually, the stored MR data signals from the separate channels are often combined to form images or spectra.
The non-MR signals can be introduced into the MR data signal path at several stages. In the wireless embodiment using transmitting aerial 5 of
It is equally possible to introduce the signal at later stages of amplification or modulation in the signal path, e.g. directly into an unused coil input port via a galvanic isolation. In principle, the signal can be introduced at any stage of the signal path within cage 11 at a frequency giving detectable changes in the digitised signal, e.g., at a low frequency immediately before digitization. Here, a cable 3 between the RF output of the MFAM and the MR apparatus will most typically be used, although it would also be possible to apply the wireless solution using dedicated coil and circuitry. As the detected MR signal is modulated from the Larmor frequency to lower frequencies along the signal path, the modulation frequency of the non-MR data signal should be adjusted accordingly depending on the point at which it is introduced.
The non-MR signal can be introduced in channels that carries also MR-signal, or it can be in separate channels. The latter option having the advantage that separation of MR and non-MR signals is inherent. As an example, several non-MR signals may be multiplexed or in other ways mixed, before being introduced via a transformer into a channel presently not used for MR-signal (even though usable for MR-signals also).
The electric connection between cable 3 and the MR data signal path is made through a transformer with galvanic isolation.
Being introduced into the MR data signal path, the non-MR data signals are processed equivalently to the MR data signals in the remaining signal path. This processing include analogue to digital conversion in ADC 15 and storage in memory 16. Analysing tools provided in relation to or implemented in the MR apparatus can be used on the raw MR data or images to demodulate and restore the non-MR signals in connection with the corresponding MR data.
In a second embodiment the functions performed in Steps III and IV are reversed and incorporated in the MR apparatus. Here, the MR apparatus comprises means for registering, measuring or monitoring non-MR signals in the scanner room during a sequence and for modulating and transmitting the non-MR signals along the MR data signal path. As for the first embodiment, analysing tools connected to or implemented in the MR apparatus can demodulate and restore the non-MR signals in connection with the corresponding MR data.
An attractive implementation of the means for registering, measuring or monitoring non-MR signals in the scanner room involve the equipment of an otherwise normal coil with one or more dedicated connections for external signal leads, e.g., sockets for EEG electrodes. The MFAM and galvanic separation from patient could be implemented in the coil, and the data would be transferred through a preferably (but not necessarily) unused channel of the scanner. Also, the input signals could be amplified or the amplifier could be implemented in the coil as well. With a special sequence treating signals from channels differently (when separate channels are used), or splitting MR- and non-MR data signals, the non-MR and MR signals could be reconstructed separately.
In another embodiment, the means comprises input ports to a combiner incorporated along the normal MR data signal path as well as MFAM and galvanic separation. An additional non-MR apparatus in the cage could be connected directly to these input ports. Here, the oscillators of the MFAM should be tuned to provide modulation frequencies suitable for the relevant stage of the signal path. Similarly, if the means includes an amplifier, the amplified signals should correspond to the equivalent MR-data signals at that stage. The non-MR signal can thereby be introduced at other positions along the paths, which may be preferable, e.g. for avoiding potentially noise-emitting equipment operating at the Larmor frequency.
Thus, using the proposed method and with minor modifications of existing coil and sequence technology, non-MR signals of interest can be measured with a normal MR apparatus during in an otherwise normal MR measuring sequence.
The modulator or MFAM can be fully or partly incorporated into the MR apparatus. In this embodiment, shown in
In the case of advanced MR-imaging it is often of interest to monitor physiological parameters of a subject during scanning. These include:
Here, the non-MR data signal is a physiological signal for monitoring the state of the subject. It is a well recognised problem that the violent changes in magnetic gradients produced by the scanner introduce signal artifacts in the physiological measurement which can normally only be overcome by precise synchronisation or retrospective filtering.
According to the state of the art, apparatus for recording physiological signals as mentioned above typically consist of a set of sensors and dedicated apparatus for amplification, sampling, analogue to digital (AD) conversion and storing. The sampling must be performed in precise synchronisation with the scanning sequence of the MR apparatus to avoid large signal artifacts in the physiological signal. Also, the sampler etc. must be MR compatible in the sense that they can tolerate the electromagnetic fields associated with MR-scanning and in that they are RF shielded in order not to generate unwanted electromagnetic fields at MR-frequencies. When the physiological signal is to be used outside the scanner room, careful filtering has to be made in order not to disturb the shielding properties of the Faraday cabin in which the scanner is located.
According to the present invention, there is no need for dedicated sampling and AD converting apparatuses when recording physiological signals.
Due to the large bandwidth of the MR apparatus, it can handle a large variety of physiological signals. Each piece of equipment for measuring a physiological signal could be equipped with its own MFAM operating at different frequencies, or they could be wired to a common MFAM which would assure that no two signals are transmitted with the same carrier frequency. Using several channels from a multi-channel MR-scanner the method could even be useful in extreme cases, such as getting the signal from hundreds of EEG-channels out of the scanner room.
The equipment feeding the physiological signals to the MFAM (e.g. electrodes and cables and possibly an amplifier) should be MR compatible in the sense that they can tolerate the electromagnetic fields associated with MR-scanning and do not generate unwanted electromagnetic fields at MR-frequencies themselves. This would typically include RF shielding of circuitry and avoidance of ferromagnetic materials in electrodes and other parts to be situated within the scanner.
Recording MR images involves a large variety of different pulse sequences which determine the types of tissue being imaged, contrast etc. Commonly used pulse sequences are Spin echo sequence, Inversion recovery sequence, Gradient echo sequences, Echo-planar pulse sequence and more. Practically, a pulse sequence consists of many short periods in succession where the MR apparatus generates different EM field conditions (magnetic and RF fields) inside the scanning part. At periods with large changes in the magnetic gradients, the signal artifact on the non-MR data signal will be very large. Therefore, it is desirable to synchronise the sampling of the non-MR data signals by the MR apparatus with periods where the magnetic gradients are constant or slowly varying.
The gradient slew rate, Gsl, is the numerical value of the time-derivative of the strength of the magnetic gradient, summed over the X, Y and Z gradients. The gradient slew rate is proportional to the expected artifact or noise in a recorded EEG signal. The artifact corresponding to each gradient component can be identified. Using a stepping stone readout gradient Gr as shown in
This technique requires microsecond-synchronisation between the fMRI sequence and the EEG sampling which can be difficult to achieve. With the MFAM, overlap between gradient switching and EEG sampling can easily be avoided by modifying the MR-sequence to only sample the MR and non-MR signal when the magnetic gradients are of constant amplitude. Most standard pulse sequence implementations will only require changes at a post-processing level to extract the non-MR signal.
After AD conversion of the combined MR and non-MR data, the raw data are stored on a hard disk or equivalent storage media. If the operator can gain access to the raw data, it is possible to design software which can extract and restore the non-MR data, provided that the measurement timing, the oscillation frequencies and the modulation scheme from the modulator (MFAM) is known. For slowly varying non-MR data, such as monitoring of breathing, it may not be necessary to access raw data to extract the non-MR signal.
Echo planar imaging (EPI) sequences are the most commonly used for functional MRI. In the following, the decoding of MR-and non-MR signals acquired simultaneously with an EPI sequence is described for the case where non-MR signals are encoded in the RF-carrier by amplitude modulation. The basic principle for extracting the measured non-MR signal is to ignore the imaging gradients as these do not influence the measured non-MR signal directly (except possibly for gradient induced artifacts). With this in mind, extraction of the non-MR signal is simple, even when gradient patters are complicated. The general principle can therefore be used for signals acquired with other imaging sequences too.
Images acquired while the MFAM is active in the scanner room appear normal except for patterns along lines orthogonal to the frequency encoding (readout) direction. Depending on the chosen modulation frequencies of the non-MR data signals, these lines may only be visible when the total sampling bandwidth is increased. With appropriate adjustment, the image of the scanned subject is undistorted. Hence, the MFAM produce very controlled image artifacts or noise signals containing reproducible data, and this artifact or noise lies nicely along the edge of the produced images without disturbing the object of the image.
Image acquisition using EPI involves high bandwidth sampling, typically around 100 kHz for approximately 100 ms. Often, the sampling period is split into shorter periods, one per line of the image. In contrast to the MR signals, the non-MR data signals emitted from the MFAM are not directly influenced by changing magnetic fields. With the chosen encoding (amplitude modulation at distinct frequencies), the encoded signals can consequently be extracted from columns of a spectrogram of the measured data. It is simple and fast to calculate a spectrogram for each image, i.e., the frequency content as a function of time. This is most easily done by organising the measured data in a matrix similar to the k-space data matrix used for image reconstruction, but with time along both axis (“true” time, and time from last beginning of a sample period). For the case of EPI, this matrix is identical to the k-space data matrix except for a reversal of every second line, and possibly an apodization. In order to calculate the spectrogram, the matrix can be Fourier transformed along the “short” time axis.
It was assumed implicitly above that the original signal and therefore the amplitude of the RF signal, does not change faster than the line time (typically 500-2000 us). This, however, is not a serious constraint, as the spectrogram can also be calculated from other groupings of the data along the two time-axis, thus allowing a trade-off of time-resolution for spectral resolution.
The squares of encoded signals can be reconstructed with, e.g., 1 ms time resolution by weighted averaging of a small number of neighbouring columns of the squared absolute spectrogram.
The outlined principles are equally valid for other imaging sequences, multi-slice imaging and imaging over longer periods of time. The encoded signals can therefore be measured simultaneously with normal imaging using the scanner. Extra sampling periods can be inserted in waiting periods to increase sampling density. Another way to increase the sampling density is to use the method outlined in example 2 in step II above: The time-displaced signals can be recombined to increase sampling density.
Since longer periods of the physiological data are acquired over many images, there are short periods during spin preparation, where no physiological signals are measured. Consequently, the frequency content needs to be estimated from uneven sampling, e.g., using gridding, Lomb-Scargle periodograms or Bayesian estimation.
Another embodiment of the invention relates to the measurement of other imaging modalities than MRI within an MR apparatus. As an example, avalanche photo diodes can be used as detectors in positron emission tomography (PET) scanning and they have been demonstrated for simultaneous PET and MRI. A practical way to integrate the PET detectors in an MR acquisition and post-processing environment is via the technique proposed in the present invention: Signals preferably derived from the coincidence detectors of a PET detector unit can be pre-processed and encoded onto one or more RF carrier signals transferred to the MR apparatus as described. This facilitates the implementation of new imaging modalities, as existing signal paths and available post-processing and visualisation tools already available in the MR environment can be employed.
In a further embodiment, the RF carrier is provided by a substance exhibiting magnetic resonance in the scanner. The non-MR data or data signal is modulated onto the magnetic field from the MR apparatus within a delimited region, where the induced MR signal is modulated and used as a carrier. This allows for very small and simple alternatives to the MFAM described in relation to
When the input terminals 47 and 48 receive an electrical signal from a variable source of electromotive force (EMF, e.g. a battery connected in series to a resistor sensitive to temperature), a corresponding current is generated in the circuit 40 including coil 42, thereby producing a magnetic field inside the coil according to Ampere's Law. The circuit 40 will thereby create a magnetic field modulated with the received electrical signal. When positioned in an MR apparatus, nuclei inside the coil 42 will experience a superposition of the magnetic fields from the MR apparatus and the modulated magnetic field from the circuit.
With proper tuning of the inductance of coil 42, the modulated MR signal from nuclei inside the coil can be extracted from the other MR data from the sequence, and can be analysed to provide the encoded non-MR data.
The modulator of
Software for extracting and restoring the non-MR data signals from the data recorded by the MR apparatus has been mentioned. The following describes a scenario for software processing of the recorded data and, given the oscillation frequencies and the modulation scheme from the modulator (MFAM), extracting the non-MR signal.
The data flow is from left to right in the figure. Starting at 90, the measured raw data contains both MR and non-MR components. The data are split for independent extraction of these components. The upper branch illustrates normal MR image reconstruction and data storage, 92 and 93. Special attention should be given to the first step 91 of limiting the raw data bandwidth: If the MFAM frequencies are chosen outside the MR frequency range, this filtering process allows the rest of the MR image reconstruction 92 to be entirely normal and to provide images free of artifacts (i.e. independent of the non-MR signals). Such bandwidth reduction corresponding to oversampling is standard signal processing employed as default for several commercial MR apparatuses.
The lower post-processing branch illustrates how extraction of the non-MR data can be performed when the MFAM frequencies have been chosen outside the frequency range of MR signals. First, 94, a spectrogram of the measured data is calculated. Based on the peaked nature of the power profile from which the MFAM frequencies can be derived, or based on priori knowledge of these frequencies, the non-MR signals can be extracted from the spectrogram and subsequently be stored (95 and 96), e.g. together with the imaging data in a PACS.
Finally, the (now separate) MR and non-MR data will often be brought together again for a correlation analysis (97 and 98) as illustrated in the right part of
A demonstration setup for measuring several EPH- and calibration signals simultaneously during echo planar MR imaging (EPI) will be presented in the following.
For this ultra-fast imaging method, the signal can be acquired during short periods with no gradient activity. In the demonstration, EPH signals have been measured with electrodes positioned near the heart, eye musculature and brain to record electrocardiogram (ECG), electrooculogram (EOG) and electroencephalogram (EEG) simultaneously. The setup is thus similar to the setup shown in
The EPH signals was measured with ECG/EEG electrodes 78 and fed to an ECG pre-amplifier 74 with high bandwidth and small settling time. The DC components from the electrodes 78 are removed from the pre-amplified signals by a high pass filter 75, and the filtered signal is fed to the gradient triggered sample hold circuit 76.
Special care is important to avoid saturation of the circuits of amplifier 1. To synchronise the sample-hold circuit 76 with the magnetic field changes, a detection coil 51 was placed within the scanner part of the MR apparatus. The induced signal was amplified by gradient activity detector amplifier 72 and monitored by gradient detection and delay circuit 73 to detect gradient activity. A high filter bandwidth of 1 MHz for the gating can ensure that also abrupt gradient changes are detected. After approximately 100 μs of silence following periods of activity (sufficient for the electrode signal to settle and to avoid the artifact from the RF pulses), the sample hold circuit 76 was opened for around 20 μs. A 10 kHz low-pass filter is associated with this, and the noise level therefore corresponds to 100 μs signal averaging per sample of the undistorted signals. The filter leaves signal at frequencies below 625 Hz essentially undistorted. In total, the delay from gradient activity to sampling was approximately 120 μs which is shorter than the period between gradient reversals even for fast EPI with an echo spacing of say 500 μs. The exact timing is not critical, as long as the amplifier has time to settle. Consequently sampling with practically any plateau-sampling EPI sequence is feasible with no adjustment of the sequence or constructed hardware.
Each channel of the EPH signals were amplified by an adjustable factor (×1600-14600 allowing input ranges of 300 μV to 3 mV, full scale) by amplifiers 74 and 77. Also, the signal was offset and scaled to ensure an always positive sign for the dynamic range of interest. This resolves sign ambiguity in subsequent demodulation steps. Artificially generated step-function signals was generated simultaneously for calibration, verification and demonstration purposes. The amplified signal and the artificially generated step-function were then sent to the MFAM 2.
The input signal was gated and low-pass filtered to avoid, channel interferences, distortion of the MR signals, and noise otherwise affecting measurements in inverse proportion to the short sample period of the sample-hold circuit. Any of the eight channels of the used MFAM can be individually switched to transmit different artificially generated step-function signals rather than measured signals. These signals generated internally in the modulator box were used for calibration, validation and demonstration purposes, but could also be used to verify, e.g., EPI timing. The channel clock and the filtered signals were fed to the amplitude modulators. The outputs from these were joined in a power combiner and sent to the RF output. The transmitter power was approximately 10 nW per channel. The aerial was a simple quarter wavelength flexible wire extending horizontally from the modulator box.
Images acquired while the modulator is active in the scanner room appear normal except for patterns along lines orthogonal to the frequency encoding (readout) direction. Depending on the chosen frequencies, these lines may only be visible when the total bandwidth is increased. With appropriate adjustment, the image of the scanned subject is undistorted.
Image acquisition using echo planar imaging involves high bandwidth sampling, typically around 100 kHz for approximately 100 ms. Often, the sampling period is split into shorter periods, one per line of the image. In contrast to the MR signals, the signals emitted from the MFAM are not directly influenced by changing magnetic fields. With the chosen encoding (amplitude modulation at distinct frequencies), the encoded signals can consequently be extracted from columns of a spectrogram of the measured data. It is simple and fast to calculate a spectrogram for each image, i.e., the frequency content as a function of time. This is most easily done by organizing the measured data in a matrix similar to the k-space data matrix used for image reconstruction, but with time along both axis (“true” time, and time from last beginning of a sample period). For the case of EPI, this matrix is identical to the k-space data matrix except for a reversal of every second line, and possibly an apodization. In order to calculate the spectrogram, the matrix is Fourier transformed along the “short” time axis.
The squares of the offset and encoded signals were reconstructed with time resolution equal to the EPI echo spacing by averaging of a small number of neighbouring columns of the squared absolute spectrogram. From these, the offset and scaled signals were recovered. Comparison to the simultaneously acquired known calibration signals can establish the original signal including sign and magnitude.
This analysis was done on a separate PC using locally developed software implemented in the programming language IDL (Research Systems Inc., Boulder, Colo., USA). The raw MR data and acquisition parameters were automatically saved on the MR acquisition computer and were copied via network to the separate PC by manually issuing a single copy command. The implemented software extracts a given number of embedded signals from the EPI raw data, and performs basic analysis. The timing information was derived from the acquisition parameters. The carrier frequencies were derived automatically from the raw data itself by analysis of the power profile averaged over all EPI readouts. The carriers may not have the most power, but since they give rise to sharp peaks in the spectral distribution, they were consistently found to have the most pronounced power difference compared to the neighbouring frequencies (pixels). Consequently, the carrier frequencies were detected as the distinct positions with the smallest second derivatives (negative) of the averaged power profile. The frequency range used for integration was not critical, but was chosen ±6 kHz (3 pixels on either side of the detected peak).
For the purpose of demonstrating the present invention technique, a simple and illustrative MRI protocol and setup was chosen. The gradient echo product sequence of the Siemens Trio®, 3 Tesla, whole-body imaging system was used for repeated echo planar imaging of three axial slices through the eye region of a healthy volunteer for 28 seconds. The standard quadrature head coil was used. In the middle third of the period, the subject paused with open eyes, whereas the remaining time was spent alternately looking right and left self-paced. This activity would normally be visible on both MR images and EOG, and the correlation between the two can therefore be used to probe the validity of the approach according to the invention.
Electrophysiological signals were therefore measured during MRI with MR compatible electrodes positioned near the heart and eye musculature to record ECG and EOG simultaneously.
The MRI parameters were as follows: Slice thickness 5 mm with 1 mm inter-slice gap. The echo and repetition times, TE/TR=41 ms/235 ms, were chosen minimal for the used 128×128 image matrix and 400 mm quadratic field of view (3 mm is a typical fMRI in-plane resolution). Oversampling in the readout-direction is automatically employed by the used scanner, thus providing extra bandwidth (only reflected in the raw data) for the encoded EPH-signals. Each EPI gradient lobe involved ramping up for 130 μs, a 300 μs plateau, and 130 μs ramping down. Sampling was performed in the middle 512 μs period, thus including 212 μs ramp-sampling, being unproblematic only due to the used gradient-activity triggering. The used parameters are representative of studies where emphasis is put on fMRI performance, i.e., the gradient artifacts in EPH-recordings are not minimized by sacrificing temporal or spatial resolution.
The EPI echo spacing was 0.56 ms which is therefore also the time resolution of the EPH measured during single-image acquisition when reconstructed as described above. Between image acquisitions, however, there are periods of excitation, spoiling and other acquisition pauses. With the used imaging parameters, there are 5 ms periods between EPI of neighbouring slices, where the EPH-signal was not measured. For a time domain-analysis, this would normally not require attention, but it must be considered if, e.g., the alpha-power of an EEG is to be estimated (discussed later).
In addition to the 128 readouts used for reconstruction of each EPI image, the used sequence acquires 3 reference lines for the purpose of ghost-correction. These are not phase-encoded. They contain encoded EPH-signals but were discarded from the EPH-signal reconstruction to avoid potential transients in the beginning of the EPI echo-train. The non-sampled interval therefore increased from 5 to 6.6 ms corresponding to EPH-signals being measured in roughly 90% of the available time.
Due to eddy-currents and other differences between sampling on positive and negative gradient lobes (sources of normal EPI-ghosting), modulation of every second EPH-sample is expected, but easily removed by filtering or by sacrificing half the bandwidth. The latter approach was chosen here, as the resulting Nyquist frequency, 446 Hz, is still sufficient for sampling of all EPH. Neighbouring samples were simply averaged giving a time resolution of 2×0.56 ms=1.12 ms except in the above-mentioned pauses between slices.
There may be additional signal variation with the EPI line number, e.g., from transient eddy currents from phase-encoding or slice selection gradients. This non-random noise can be estimated and filtered relatively easily but this was not necessary for the present application. It is likely to be necessary for EEG recordings.
To avoid possible modulation with the slice number, e.g., coming from MR signal leakage into the EPH-signal region, the data were filtered in the following way: The extracted EPH-signals were averaged over each image, so the time resolution was temporarily decreased from 0.56 ms to TR/3=78 ms equidistant sampling. With one sample per image, it is simple to remove oscillation in synchrony with the slice excitation by use of a filter exchanging the spectral content at frequencies ±1/(TR) with the means of the neighbouring frequencies differing by the reciprocal of the total scanning time. Subsequently, the low-frequency part of the original high-resolution time series was exchanged with the filtered version, thus removing slice selection effects occurring at single sharp frequencies in a simple and effective way. For recording of rhythmic electrophysiological signals for short periods of time, the timing or the filtering approach may need to be reconsidered, if modulation is observed.
The MR images were reconstructed using 2D Fourier transformation after Gaussian smoothing in the readout direction (2 pixel full width at half maximum, sometimes used in MRI analysis). This suppresses phase discontinuities at the edges of the acquisition matrix, thus effectively preventing side lobes of the encoded signal to leak into the MRI object region. Alternatively, the additional RF signal could be estimated and subtracted relatively easily, as it has near constant amplitude in each sub-millisecond readout period.
The positioning of the amplifier and modulator in the MR room proved to be easy and non-critical. Several positions were used, and no special attention to the orientation or positioning of the antenna was needed, although those parameters influence the amplitude of the measured non-MR signals. As the signals are influenced equally, the sampling of the known calibration signal provides a reference that can be used to make quantitative measurements.
The simple data extraction algorithm worked reliably, even when RF carriers were weak, or when they were overlapping with MR frequencies (in which case, of course, both MR images and EPH-recordings are corrupted). Consequently, for shorter acquisition periods (less than a minute) the MR images and the EPH-recordings could be viewed and correlated seconds after the acquisitions were finished with the number of EPH-signals being the only parameters needing to be manually specified. For longer acquisitions, the network copying was the most time-consuming step. The following graphs are output from the software.
The spectrogram of
High correlation between the EOG and the image intensity is found in the eye region, and similarly, correlation is seen between the ECG and the image intensity near the medial cerebral artery and the circle of Willis, demonstrating the validity of the approach. High correlation is also found in other regions, most notably outside the head, as the signals are encoded as artifacts there. The signals are seen to take up more bandwidth than needed and the overall noise in the MR images is therefore increased. In contrast to
The physiological signals are measured via MR imaging and their sampling is therefore normally non-equidistant. Appropriate methods must be used for calculating the spectrogram. In the graph of
| Number | Date | Country | Kind |
|---|---|---|---|
| PA 2004 00818 | May 2004 | DK | national |
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/DK2005/000343 | 5/25/2005 | WO | 00 | 3/18/2008 |
