Multichannel signal enhancement

Claim 2

A mixed adaptive filter for generating enhanced signal data from successive sensor input data using a known signal-to-data cross correlation, comprising: first processor means connected in a first triangular array for rotating second sensor input data into a first stored triangular matrix functionally related to first sensor input data to form a second triangular matrix and having an output column for outputting a first intermediate output; second processor means connected in a second triangular array for back substituting said first intermediate output into said second triangular matrix to form a second intermediate output; and third processor means for combining said second intermediate output with said known signal-to-data cross correlation to output said enhanced signal data.

Claim 3

A filter according to claim 1, wherein said first and second triangular arrays are congruent for simultaneous processing using said second triangular matrix.

Claim 4

A mixed adaptive filter for generating enhanced signal data from successive sensor input data using a known signal-to-data cross correlation, comprising: first processor means connected in a first triangular array for rotating second sensor input data into a first stored triangular matrix functionally related to first sensor input data to form a second triangular matrix and having an output column for outputting a first intermediate output; second processor means connected in a second triangular array for forward elimination of said second triangular matrix with said known signal-to-data cross correlation to form a second intermediate output; and third processor means for multiplying said first intermediate output with said second intermediate output to form said enhanced signal data.

Claim 5

A mixed adaptive filter for generating enhanced signal data from successive sensor input data using a known signal-to-data cross correlation, comprising: first processor means connected in a first triangular array for rotating second sensor input data into a first stored triangular matrix functionally related to first sensor input data to form a second triangular matrix and having an output column for outputting a first intermediate output; second processor means connected in a second triangular array for forward elimination of said second triangular matrix with said known signal-to-data cross correlation to form a second intermediate output; and third processor means for multiplying said first intermediate output with said second intermediate output to form said enhanced signal data.

Claim 6

A filter according to claim 3, wherein said first and second triangular arrays are congruent for simultaneous processing using said second triangular matrix.

Claim 7

An MEG signal processing filter for generating an enhanced output signal from an estimated noise autocorrelation and input data obtained in a single trial run having successive signal inputs, including an evoked response from a selected stimulus, comprising: first processor means connected in a first triangular array for storing a first triangular signal matrix from first signal data and for generating successive stored updated triangular matrices by rotating second data into said first triangular matrix and rotating successive signal data into corresponding ones of said stored updated triangular matrices and having a column output for providing successive first intermediate outputs; second processor means connected in a second triangular array for back substituting each said first intermediate output with a corresponding updated triangular matrix to form successive second intermediate outputs; and third processor means for combining said second intermediate outputs with said estimated noises autocorrelation to form said enhanced MEG signal.

Claim 8

An MEG signal processing filter for generating an enhanced output signal from an estimated noise autocorrelation and input data obtained in a single trial run having successive signal inputs, including an evoked response from a selected stimulus, comprising: first processor means connected in a first triangular array for storing a first triangular signal matrix from first signal data and for generating successive stored updated triangular matrices by rotating second data into said first triangular matrix and rotating successive signal data into corresponding ones of said stored updated triangular matrices and having a column output for providing successive first intermediate outputs; second processor means connected in a second triangular array for back substituting each said first intermediate output with a corresponding updated triangular matrix to form successive second intermediate outputs; and third processor means for combining said second intermediate outputs with said estimated noises autocorrelation to form said enhanced MEG signal.

Claim 9

An MEG signal processing filter according to claim 5, wherein said first and second triangular arrays are congruent for simultaneous processing involving each said updated matrix.

Claim 10

A process for generating an enhanced MEG output signal from sensor input data obtained in a single trial run including an evoked response from a selected stimulus, comprising the steps of: obtaining first sensor data; generating an initial triangular matrix and an autocorrelation matrix from said first sensor data; storing said initial triangular matrix in a corresponding triangular array of data processors; obtaining second sensor data during said selected stimulus; processing said second sensor data through said triangular array in a first direction by updating said initial triangular matrix to form successive triangular matrices by successive rotations and to output successive first intermediate outputs; successively processing said first intermediate outputs by back substitution with a corresponding one of said successive triangular matrices in a second direction through said triangular array to form successive second intermediate outputs; and converting said successive second intermediate outputs with said autocorrelation matrix to form said enhanced MEG output signal.

Claim 11

A process for generating an enhanced MEG output signal from sensor input data obtained in a single trial run including an evoked response from a selected stimulus, comprising the steps of: obtaining first sensor data; generating an initial triangular matrix and an autocorrelation matrix from said first sensor data; storing said initial triangular matrix in a corresponding triangular array of data processors; obtaining second sensor data during said selected stimulus; processing said second sensor data through said triangular array in a first direction by updating said initial triangular matrix to form successive triangular matrices by successive rotations and to output successive first intermediate outputs; successively processing said first intermediate outputs by back substitution with a corresponding one of said successive triangular matrices in a second direction through said triangular array to form successive second intermediate outputs; and converting said successive second intermediate outputs with said autocorrelation matrix to form said enhanced MEG output signal.

Claim 12

A process according to claim 7, further including the steps of: obtaining said first sensor data in the absence of said selected stimulus to form a noise autocorrelation matrix; and converting said successive second intermediate outputs with said noise autocorrelation matrix and subtracting said second sensor data to form said enhanced MEG output signal.

Claim 13

A process according to claim 7, further including the steps of: obtaining said first sensor data in a plurality of trial runs including said stimulus; generating a time-varying signal autocorrelation matrix by averaging said first sensor data; and converting said successive second intermediate outputs with said signal autocorrelation matrix to form said enhanced MEG output signal.