Composite brain function monitoring and display system

Inventors

Liley, David Tibor Julian • Sinclair, Nicholas Campbell

Abstract

A method of displaying the activity of a brain, the method including the steps of: (i) obtaining an electroencephalogram (EEG) signal from the brain; (ii) segmenting said EEG signal into either contiguous or overlapping segments comprised of a sequential number of samples of said EEG signal; (iii) representing said EEG segments as a fixed order autoregressive moving average (ARMA) signal representation with an autoregressive order between 8 and 13 and a moving average order between 5 and 11; (iv) rewriting in z-domain notation said fixed order ARMA signal representation to obtain a z-domain representation; (v) generating AR coefficient data and MA coefficient data for said segments of said EEG signal for said fixed order ARMA signal representation: (vi) determining the poles and zeros of said z-domain representation for said segments of said EEG signal by substituting said coefficient data for said segments of said EEG signal into said z-domain representation; (vii) calculating the sum of the number of poles determined in step (vi); (viii) calculating the sum of the number of zeros determined in step (vi); (ix) representing said ARMA signal representation as an infinite order autoregressive (AR) model in z-domain notation; (x) determining autoregressive coefficient data for said infinite AR model from the AR and MA coefficient data generated in step (v) of said fixed order ARMA representation; (xi) determining the sum of the poles for said infinite order AR model for said segments of said EEG signal as either: (a) the first autoregressive coefficient of said finite order AR model; or (b) the difference of the sum of poles and the sum of zeros as determined respectively in steps (vii) and (viii); (xii) determining an index value representing the activity of the brain for said segments of said EEG signal by applying a discriminating function to the sum of the poles of said infinite AR model for said segments of said EEG signal as determined in step (xi); and (xiii) displaying said index value on display means.

Core Innovation

The invention relates to monitoring and displaying the activity of a brain by obtaining an electroencephalogram (EEG) signal using a plurality of electrodes and digitising means. The EEG signal is segmented into either contiguous or overlapping segments comprised of a sequential number of samples, and each segment is processed for display of an index value representing brain activity.

Each EEG segment is represented as a fixed order autoregressive moving average (ARMA) signal representation with an autoregressive order between 8 and 13 and a moving average order between 5 and 11. The fixed order ARMA representation is rewritten in z-domain notation to obtain a z-domain representation, and autoregressive coefficient data and moving average coefficient data are generated for the segment. The poles and zeros of the z-domain representation are determined by substituting the coefficient data into the z-domain representation.

From the poles and zeros, the ARMA signal representation is represented as an infinite order autoregressive (AR) model in z-domain notation. Autoregressive coefficient data for the infinite AR model is determined from the AR and MA coefficient data, and the sum of poles for the infinite AR model is determined either as the first autoregressive coefficient or as the difference of the sum of poles and the sum of zeros. An index value representing the activity of the brain is generated by applying a discriminating function to the sum of the poles of the infinite AR model, and the index value is displayed on display means.

In additional embodiments, the cortical state and composite cortical state are generated from pole sums, and a discriminating function can be obtained via discriminant analysis to convert pole-based measures into the index value. The invention also includes generating cortical input measures derived from an ARMA gain and a discriminating function to combine poles with cortical state and cortical input measures for the index value. Display concepts include graphs over time and color-coded predetermined value ranges for the index value.

Claims Coverage

The independent claims identified are clm-00001 and clm-00012. They cover the core processing chain that converts segmented EEG into a fixed-order ARMA representation, derives poles and zeros via z-domain rewriting, converts to an infinite-order AR model, computes a sum of poles, applies a discriminating function to produce an index value, and displays the index value. The inventive features include ARMA order ranges and the discriminating-function-based pole-sum index.

Across clm-00001 and clm-00012, the claims cover a discriminating-function-based brain activity index generated from a pole sum derived from segmented EEG converted from a fixed-order ARMA model with specified AR/MA order ranges through z-domain pole and zero calculation into an infinite-order AR model, followed by display of the index value.

Stated Advantages

Documented Applications