Identifcação do padrão de hipsarritmia em eletroencefalogramas: utilizando decomposição de sinais em pequenas ondas

Abstract

Epilepsy is a temporary change in brain function.It is possible during an epileptic seizure to see marked changes. In infants, for example, some encephalopathies such as West syndrome stand out. West syndrome is mainly characterized by infantile spasms and hipsarrhythmia. Hypsarrhythmia, first described in 1952, is characterized by an electroencephalographic pattern composed of slow waves and high-voltage anarchic projection spicules. The pattern of hipsarrhythmia is also present in EEG's of children with microcephaly caused by Zika virus. Although the characteristics of hipsarrhythmia in the electroencephalogram are well defined, the identification of this pattern still causes disagreement among specialists. This work proposes the development of mathematical and computational methods capable of identifying the basic characteristics of this pathological electrical signal, in order to assist in the diagnosis and prognosis of epileptic patients. During this work we present the mathematical formulation of the continuous and discrete Wavelet Transform of the Gabor functions that will be used as the core of this transformation and the tuned Gabor functions. The calculation of the three indices in electroencephalogram signals of infants with Zika virus was applied to identify hipsarrhythmia. These signals were acquired from the Maranhão State Government project called Casa Ninar. The results may show the effectiveness of the methodology in identifying the pattern of hipsarrhythmia, since they present rates of accuracy that range from 95 to 100 % in the classification of electroencephalograms. In the most difficult task, which is to classify sections of the EEG, the hit rates ranged from 75 to 90 %.