CLASSIFICATION OF SINGLE-TRIAL ERP SUBTYPES - APPLICATION OF GLOBALLYOPTIMAL VECTOR QUANTIZATION USING SIMULATED ANNEALING

Examination of the single trials which are traditionally averaged to f orm late-component ERPs reveals a number of different sub-types of res ponse. This study introduces an automated and robust approach to objec tively classify these ERP sub-types. Auditory oddball ERP (target tone s) data were examined in 25 normal subjects. Globally optimal vector q uantization using simulated annealing (the ''Metropolis algorithm'') w as employed to determine the natural groupings of the single-trial res ponses that constitute the average ERP. No prior assumptions about the ERP patterns were imposed. This is the first study to employ a cluste r analysis technique with globally optimal properties in ERP research. We demonstrate that, due to the presence of many different undesirabl e local minima, a globally optimal solution is crucial if the classifi cation of the single-trial ERPs is to reflect their real structure. Th e results of this study showed that only around 40% of single trials h ad a morphology which resembled the averaged ERP wave form. The remain ing single trials had a response morphology which was different from t he average, in terms of the amplitude and latency of the components. S ingle-trial ERP response sub-types may provide fundamental complementa ry functional information to the ERP average.