Automatized identification of absence-like seizures in electrocorticogram of rats

Abstract

Absence-like seizures are presented in rat electrocorticogram as paroxistics spike-wave discharges. The aim of this study was to develop and test an automatized system for seizures identification. System fundament is to distinguish seizure signals from others by power analysis in frequency range between 7 and 11 Hz, which is high only during seizures. Routine, elaborated at Matlab®, is inputted with an interval without seizures (sample) in order to set up the criterion of identification. The system compares, through the Fourier fast transformed, the sample with the entire recording. Program output is a table that reports the presence of seizures in such intervals. Four Wistar rats genetically prone to show absence-like seizures with implanted cortical electrodes were submitted to 80 minutes of electrocorticogram. Seizures were counted manually minute by minute. Recording and seizure quantification of one animal were used to refine the system. Automatized quantification was very similar to manual quantification for the animal used as sample. Recordings of the other animals, used to test the system, shows similar pattern. The system is able to quantify with high fidelity the spike-wave discharges, so that it is a powerful instrument to analyze electroencephalogram in studies with animal models of this pathology. Additionally, minor changes allow identifying other patterns found in electrophysiological recordings (sleep, wakefulness, pathological manifestations etc). The routine is available for download at www.ib.usp.br/~rpavao.