SPIKE DETECTION .1. CORRELATION AND RELIABILITY OF HUMAN EXPERTS

A panel of 5 experienced electroencephalographers detected spikes in E EG trials from 40 epilepsy patients and 10 control subjects. 1952 spik es were detected, and detailed attribute scores were recorded. Statist ics from the theory of measurement error are utilized to quantify the reliability and difficulty of the study. An extension of the Pearson c orrelation coefficient, called the detection correlation coefficient, is derived and used in recognition of the fact that the readers agree on numerous non-spike regions. Spike perception is modeled with both d ichotomous and continuous values. As expected, the study reliability i s higher when using the continuous values. Standard sensitivity and sp ecificity definitions are extended and applied to continuous-valued sp ike perception. A database of ''panel scores'' was created from the 5 reader scorings by merging spikes within 75 msec on each side. The ave rage inter-reader correlation is 0.79 with a corresponding reliability of 0.95. Average spike attributes are calculated, and the resulting d atabase can serve as a ''gold standard'' for testing computer algorith ms or other readers.