EEG spatial filter and method

• US 4,753,246 A
• Filed: 03/28/1986
• Issued: 06/28/1988
• Est. Priority Date: 03/28/1986
• Status: Expired due to Term

• Current Assignee: Regents of the University of California (University of California)
• Original Assignee: Regents of the University of California (University of California)
• Inventors: Freeman, Walter J.
• Primary Examiner(s): Kamm, William E.
• Assistant Examiner(s): MANUEL, GEORGE C
• Application Number:
  US06/845,564
• Time in Patent Office: 823 Days
• Field of Search: 128/731, 128/732, 364/417, 364/724
• US Class Current: 600/544
• CPC Class Codes: A61B 5/374 Detecting the frequency distribution of signals, e.g. detecting delta, theta, alpha, beta or gamma waves

Abstract

Method and apparatus for producing enhanced EEG or MEG information related to a selected brain activity in a subject. The apparatus includes a two-dimensional array of at least about 16 sensors for recording EEG or MEG traces from the subject. Control and test traces recorded before and during an interval in which the brain activity is occurring, respectively, are each decomposed into a series of basis functions which may be analytic components such as temporal frequency components, generated by spectral decomposition of an ensemble average of the recorded traces, or principal components determined by principal component analysis. The control and test traces are then represented as a sum of the products of the individual basis functions times a spatial domain matrix which represents the spatial pattern of amplitudes of that basis function, as measured by the individual sensors in the array. The signals can be extracted by filtering spatially and/or temporally to remove basis function components which are not related to the selected brain activity (clutter), and to remove spatial frequencies inherent in the spatial domain matrices to optimize the contrast between control and and corresponding test matrices, for each selected basis function.

Claims

1. A machine-implemented method for enhancing EEG or MEG information related to a selected behavioral activity in a subject, comprising

attaching to the subject, a two-dimensional array of at least about 16 sensors for recording EEG or MEG traces from the subject, where the sensors are spaced from one another a distance which is substantially no greater than one-half the shortest expected spatial wavelength which is related to the selected behavioral activity, and where the area of placement of the array is one in which signals related to the selected activity can be detected by each sensor in the array,

recording test EEG or MEG traces by the sensors over a period in which the selected behavior is manifested,

selecting a test time interval within the recording period when such activity is occurring;

decomposing the recorded traces into a series of basis functions B_i, each of which represents a signal or clutter component which is present to various degrees in all of the traces recorded by the array sensors in the selected test interval,

determining, for at least one signal basis function B_j that is related to the selected behavioral activity, an amplitude matrix A_j which represents the spatial pattern of amplitudes of that basis function, as measured by the individual sensors in the array, in the selected test interval, and

representing test signals S_a recorded during the test interval by the array of sensors in the form;

space="preserve" listing-type="equation">S.sub.a =Σ

.sub.j A.sub.j B.sub.j (basis functions).

2. The method of claim 1, wherein the two-dimensional array includes a substantially rectangular array of at least about 8×

8 sensors.

3. The method of claim 1, wherein the modal spacing between adjacent sensors is between about 0.5 mm and 2.0 cm.

4. The method of claim 1, wherein selecting the test time interval includes the steps of (a) constructing an ensemble average of the recorded signals over the recording period, (b) constructing a moving time window within the recording period, (c) determining for each position of the time window, the ratio of average signal amplitude to the spatial standard deviation of the ensemble average, and (d) selecting each time window which shows a sustained maximum or near-maximum value of such ratio.

5. The method of claim 1, wherein said decomposing includes the steps of (a) constructing an ensemble average of the recorded traces over the selected time interval and (b) performing a spectral decomposition of the ensemble average trace to obtain a series of such basis functions B_i which represent temporal frequency components of the ensemble average.

6. The method of claim 5, wherein performing the spectral decomposition includes (a) determining a first basis function B₁ representing the dominant frequency component in the ensemble average and having estimated frequency, phase, and amplitude values, (b) subtracting the B₁ basis function frequency component from the ensemble average to obtain a residual ensemble average, and (c) repeating the determining and subtracting steps to obtain a series of basis functions B₂ -B_n representing successively less dominant frequency components, until the residual ensemble average remaining after removing the B_n basis function contains about 10% or less of the total energy of the original ensemble average.

7. The method of claim 1, wherein said decomposing includes using principal component analysis to generate a series of basis functions B_j which represent the dominant components of signal values over the selected time interval.

8. The method of claim 1, wherein the signal basis functions B_j which are related to the selected brain state are determined by the steps of:

recording control EEG or MEG traces by the sensors during a control time interval which occurs either before or after such brain activity,

decomposing the control traces into a series of basis functions B'"'"'_i, each of which (a) represents a signal component which is present to various degrees in all of the traces recorded by the array sensors in the control time interval, and (b) corresponds to a B_i basis function in the test signals,

determining, for each B_i basis and corresponding B'"'"'_i basis function, an associated A_i and C_i amplitude matrix, respectively, which represents the spatial pattern of amplitudes of the associated basis function, as measured by the individual sensors in the array during the test and control intervals, respectively, and

selecting from among the basis functions B_i, those signal basis function B_j for which the associated amplitude matrix A_j shows a statistically meaningful variance with the corresponding amplitude matrix C_j.

9. The method of claim 1, which further includes operating on one or more of the A_j amplitude matrices to remove spatial frequencies which are substantially unrelated to the selected behavioral activity.

10. The method of claim 9, wherein said operating includes transforming the amplitude matrix to the spatial frequency domain, generating a spatial frequency matrix, filtering the spatial frequency matrix to remove spatial frequencies which are not associated with the selected brain activity, and transforming the filtered spatial frequency matrix back to the spatial domain, or performing the equivalent convolution on the spatial-distribution matrix.

11. The method of claim 9, wherein the spatial frequencies which are substantially unrelated to the selected behavioral activity are identified by the steps of:

recording control EEG or MEG traces by the sensors during a control interval which occurs either before or after such brain activity,

decomposing the control traces into a series of basis functions B'"'"'_i, each of which (a) represents a signal or clutter component which is present to various degrees in all of the traces recorded by the array sensors in the control time interval, and (b) corresponds to a B_i basis function in the test traces,

determining, for each B_i and corresponding B'"'"'_i basis function , an A_i and C_i amplitude matrix, respectively, which represents the spatial pattern of amplitudes of the associated basis function, as measured by the individual sensors in the array dur-ing the test and control intervals, respectively, and

measuring the degree of statistical correlation between the A_i and corresponding C_i matrices as a function of spatial frequency cutoff, to identify those spatial frequencies which optimize the contrast between the two matrices.

12. The method of claim 1, for recovering a spatial pattern of cell-generated signals within the brain of the subject which are related to the selected behavioral activity, and which are responsible for the traces recorded on the subject by the array of sensors, which further includes computing a transform function which relates the enhanced signal S_a =Σ

_j A_j B_j (basis functions) to signal generation within the brain during the selected brain activity, according to the laws of conduction in a volume, and computing the product of the signal S_a and the inverse transform function, to generate such spatial pattern of cell-generated signals.

13. A machine-implemented method for enhancing EEG or MEG information related to a selected brain activity in a subject, comprising:

attaching to the subject, a two dimensional array of at least about 16 sensors for recording EEG or MEG traces from the subject, where the sensors are spaced from one another a distance which is substantially no greater than one-half the shortest expected spatial wavelength which is related to the selected behavioral activity, and where the area of placement of the array is one in which signals related to the selected event can be detected by each sensor in the array,

recording control and test EEG or MEG traces by the sensors during control and test time intervals, respectively, which occur before and during such selected brain activity, respectively,

decomposing the recorded test and control signals into a series of basis functions B_i and B'"'"'_i, respectively, each of which represents a signal or clutter component which is present to various degrees in the test and control traces, respectively, recorded by the array sensors in the selected test or control intervals,

determining, for each B_i and corresponding B'"'"'_i basis function , an A_i and C_i amplitude matrix, respectively, which represents the spatial pattern of amplitudes of the associated basis function, as measured by the individual sensors in the array during the test and control intervals, respectively,

selecting from among the basis functions B_i, those signal basis functions B_j for which the associated amplitude matrix A_j shows a statistically meaningful variance with the corresponding amplitude matrix C_j.

representing test signals S_a recorded by the sensors in the form;

space="preserve" listing-type="equation">S.sub.a =Σ

.sub.j A.sub.j B.sub.j (basis functions).

14. The method of claim 13, which further includes measuring the degree of statistically correlation between the A_j and corresponding C_j matrices as a function of spatial frequency cutoff, to identify those spatial frequencies which reduce the contrast between the two corresponding matrices, filtering the A_j matrices to remove the identified spatial frequencies, producing the corresponding filtered matrix FA_j, and representing test signals S_a recorded by the electrodes in the form:

space="preserve" listing-type="equation">S.sub.a =Σ

.sub.j FA.sub.j B.sub.j (basis functions).

15. An apparatus for providing enhanced EEG or MEG information related to a selected behavioral activity in a subject, comprising:

a two-dimensional array of at least about 16 sensors for recording EEG or MEG traces from the subject, where the sensors are spaced from one another by a distance which is substantially no greater than one-half the shortest expected spatial wavelength which is related to the selected behavioral activity, and where the area of placement of the array is one in which signals related to the selected event can be detected by each sensor in the array,

means for recording test EEG or MEG traces by the sensors during a test interval in which the selected brain activity is occurring, and

means for filtering the recorded traces to remove clutter components which are unrelated to such behavioral activity, said filtering means including means for decomposing the test traces into a series of basis functions B_i, each of which represents a signal or clutter component which is present to various degrees in all of the signals recorded by the array sensors in the test interval, and means for determining, for at least one signal basis function B_j that is related to the selected brain activity, an amplitude matrix A_j which represents the spatial pattern of amplitudes of that basis function, as measured by the individual sensors in the array, thereby to generate filtered signals recorded during the test interval by the array of-sensors in the form;

space="preserve" listing-type="equation">S.sub.a =Σ

.sub.j A.sub.j B.sub.j (basis functions).

16. The apparatus of claim 15, which is used to record traces both before and during the selected behavioral activity, and to generate both control and test basis functions B_i, B'"'"'_i, respectively, and associated spatial-pattern matrices A_i and C_i, which further includes means for measuring the extent of statistical variance between C_i and the corresponding A_i, to determine which of the B_i basis functions is related to the selected brain activity signal and which is not related and is clutter.

17. The apparatus of claim 16, which further includes means for spatially filtering A_i to enhance the contrast between A_i and the corresponding C_i.

18. The apparatus of claim 17, for use in recovering a spatial pattern of cell-generated signals within the brain of the subject which are related to the selected behavioral activity, and which are responsible for the signals recorded on the subject by the array of sensors, which further includes means for computing a transform function which relates the enhanced signal S_a =Σ

_j A_j B_j (basis functions) to signal generation within the brain during the selected brain activity, according to the laws of conduction in a volume, and means for computing the product of the inverse of the signal S_a and the inverse transform function, to generate such spatial pattern of cell-generated signals.