Method and system for detecting and removing EEG artifacts
A method and system for detecting and removing EEG artifacts is disclosed herein. Each source of a plurality of sources for an EEG signal is separated for a selected artifact type. Each source of the plurality of sources is reconstituted into a recorded montage and an optimal reference montage for recognizing the selected artifact type of each source. The sources with artifacts are removed and the remaining sources are reconstituted into a filtered montage for the EEG signal.
1. A method for processing an electroencephalogram (“EEG”) recording to generate a filtered montage for an EEG signal, the method comprising:
separating, by a processor, an epoch of an original EEG recording using a blind source separation to generate a plurality of sources for the original EEG recording wherein a sign and physical amplitude of a source signal of the plurality of sources is unknown, each of the plurality of sources separated for an eye blink artifact removal, the original EEG recording generated by an EEG machine component comprising a CPU, an amplifier component, and a plurality of electrodes attached to a patient's head, wherein the plurality of electrodes comprises 10-20 electrodes with a CZ reference electrode;
reconstituting, by the processor, a first source of the plurality of sources to generate a recorded montage and an optimal reference montage for recognizing the eye blink artifact type;
examining, by the processor using a neural network, a plurality of channels of the optimal reference montage for the first source to determine if the eye blink artifact type is the true artifact of the first source;
removing, by the processor, the first source of the plurality of sources for the eye blink artifact type;
reconstituting, by the processor, a second source of the plurality of sources into a recorded montage and an optimal reference montage for recognizing a muscle artifact type;
examining, by the processor using the neural network, a plurality of channels of the optimal reference montage for the second source to determine if the muscle artifact type is the true artifact of the second source;
removing, by the processor utilizing blind source separation-canonical correlation analysis, the second source for the muscle artifact type;
reconstituting, by the processor, a plurality of remaining sources into a filtered montage for the EEG signal; and
displaying the filtered montage for the EEG signal on a graphical user display for a clearer illustration of a brain activity of the EEG signal than the original EEG signal wherein the muscle artifact type and the eye-blink artifact type are removed in the filtered montage.