THE AIM OF THIS PROJECT IS TO DEVELOP SYSTEM IDENTIFICATION AND NOISE CANCELLATION SOFTWARE TO IMPROVE THE ROBUSTNESS OF MAGNETOENCEPHALOGRAM (MEG) RECORDINGS IN NONSHIELDED ENVIRONMENTS.

THE AIM OF THIS PROJECT IS TO DEVELOP SYSTEM IDENTIFICATION AND NOISE CANCELLATION SOFTWARE TO IMPROVE THE ROBUSTNESS OF MAGNETOENCEPHALOGRAM (MEG) RECORDINGS IN NONSHIELDED ENVIRONMENTS. THE MEG HAS GREAT CLINICAL POTENTIAL (E.G., IN LOCALIZATION OF SEIZURE ORIGIN), BUT THE RECORDED SIGNAL IS MANY ORDERS OF MAGNITUDE LOWER IN AMPLITUDE THAN THE ENVIRONMENTAL MAGNETIC FIELDS, AND THE LOW SIGNAL-TO-NOISE RATIO REQUIRES AVERAGING OF MANY EVENTS FOR LOCALIZATION OF SOURCES. MANY MEANS HAVE BEEN ADOPTED TO INCREASE THE SNR (PARTUCULARLY BAD FOR LOW FREQUENCIES, E.G., BELOW 2 HZ), INCLUDING USE OF SECOND-ORDER GRADIOMETERS AND THE USE OF EXTRA NOISE REFERENCE CHANNELS. AS SIGNIFICANT COMPONENTS OF THE "NOISE" CHANGE CHARACTER OVER TIME, SIMPLE SUBTRAC- TION SCHEMES IN WHICH A FIXED PORTION OF THE NOISE REFERENCEIS SUBTRACTED FROM THE MEG ARE NOT OPTIMAL. THEREFORE, ALGORITHMS FOR ADAPTIVE NOISE CANCELLATION, WHICH HAVE BEEN SUCCESSFULLY USED IN COMMUNICATIONS PROBLEMS AS WELL AS FOR REMOVAL OF EYE MOVEMENT ARTIFACT FROM EEG RECORDINGS, WILL BE APPLIED TO BRAIN MAGNETIC SIGNALS. WHEN INCORPORATED IN EFFICIENT, ROBUST, AND PORTABLE SOFTWARE, THESE METHODS WILLBE AN IMPORTANT ELEMENT IN CLINICAL AND RESEARCH APPLICATIONS OF THE MEG.