Automation of seizure duration estimation during ECT; use of fractal dimension

Obtaining seizure of an adequate length is essential during electroconvulsive therapy (ECT). Here, the authors describe an automated procedure of estimating the seizure duration using running fractal dimension. Fractal dimension (FD) or EEG records from 20 patients (14 males, age range 15-45 yrs) receiving ECT was calculated. The lowest FD values occurred in the post-seizure phase. Running fractal dimension on epochs of 4 seconds length each with 75% overlap was also computed. A computer algorithm was developed to detect the EEG epoch with lowest FD and the seizure duration was estimated thereof. Seizure duration estimates from automated method (mean, SD=67.95, 31.7 secs) agreed significantly (intraclass correlation, `r´=0.9, p<0.01) with those of clinical rating of analog EEG records (mean, SD=69.15, 32.2 secs)