Automatic pattern recognition of epileptiform discharges using morphological descriptors and linear discriminant analysis

This paper presents the performance analysis of a methodology for automated recognition of epileptiform patterns using morphological descriptors and Linear Discriminant Analysis. Morphological descriptors, in this paper, are parameters related to the morphology of the signal's waveform and Linear Discriminant Analysis (DA) is a method of multivariate statistical analysis commonly used for classification, size reduction and/or feature extraction. Thus, the main purpose of this paper is to analyze the classification performance of the discriminant functions and examine the applicability of Discriminant Analysis in reducing the number of independent variables (in this case morphological descriptors) necessary to obtain a discriminant function with acceptable classification performance. Simulations showed that the best functions exhibited efficiency greater than or equal to 85%, sensitivity of 85-90% and specificity between 80 and 84%.