Power Spectrum and Data Clustering Analysis for Intraoperative EEG Signals
Beni R. Hermanto *
Department of Biomedical Engineering, Bandung Institute of Technology, Bandung, Indonesia.
Ahmad Faried
Department of Neurosurgery, Faculty of Medicine, Universitas Padjadjaran−Dr. Hasan Sadikin Hospital, Bandung, Indonesia.
Agung B. Sutiono
Department of Neurosurgery, Faculty of Medicine, Universitas Padjadjaran−Dr. Hasan Sadikin Hospital, Bandung, Indonesia.
Muhammad Z. Arifin
Department of Neurosurgery, Faculty of Medicine, Universitas Padjadjaran−Dr. Hasan Sadikin Hospital, Bandung, Indonesia.
Richard Mengko
Department of Biomedical Engineering, Bandung Institute of Technology, Bandung, Indonesia.
Tati L. E. Rajab
Department of Biomedical Engineering, Bandung Institute of Technology, Bandung, Indonesia.
*Author to whom correspondence should be addressed.
Abstract
Intraoperative EEG is used for acquiring brain signal that probes or electrodes placed on brain organ directly. It is different from common EEG, which probes placed on scalp. In order to explore the characteristic of brain signal based on brain injuries case, data taken from ten subjects while applied intraoperative EEG. The signals acquire by placing eight channels on brain organ simultaneously with particular form of probes.
For comparing the brain signal among the subjects, power spectrum chosen as basic method. The power spectrum indicates the energy of signals, representing the brain activity. Cross checking between powers spectrum and brain injuries case, data clustering applied using self-organizing maps.
Calculating the power spectrum of signals shows that brain stroke case has higher value than non-stroke case. This higher value exists for most of channels. Using channels as dimension of data, self-organizing maps visualize that stroke case’s position are closed to each other on map. On map also, visualize the boundary between stroke and non-stroke case. Based on brain injuries happened among the subjects, stroke case has specific signals characteristic, which different from non-stroke significantly.
Keywords: Intraoperative electroencephalograph (EEG), power spectrum, fast fourier transform (FFT), data clustering