Procedure to classify field patterns.

摘要

In a procedure for classifying field patterns which are generated by electrophysiological activities (8) occurring in the interior of a living being (4) and are measured by means of a multi-channel measuring arrangement (12) outside the living being (4), a trainable classifier (34) is connected to the multi-channel measuring arrangement (12). The trainable classifier (34) exhibits a number of input channels (32) which are supplied with attribute vectors which are representative of the measured field patterns. The trainable classifier (34) is trained with training-field patterns which have been generated by a locatable equivalent model of the electrophysiological activity (8). A probability value which specifies the probability with which each field pattern can be generated by the locatable equivalent model is output for each field pattern at an output channel (38).