Hardware Design of Seizure Detection Based on Wavelet Transform and Sample Entropy

Abstract

Detecting epileptic seizure is a very time consuming and costly task if a support vector machine (SVM) hardware processor is used. In this paper, an automated seizure detection scheme is developed by combining discrete wavelet transform (DWT), sample entropy (SampEn) and a novel classification algorithm based on each wavelet coefficient and voting strategy. In order to save circuit area, a Daubechies order 4 (db4) filter of lattice structure is introduced in DWT, only half elements of the symmetric distance matrix in the SampEn are stored and module reusing strategy is used. To speed up the detection, intermediate results are reused by reasonably organizing the SampEn calculation procedures. The seizure detection scheme is implemented in a field-programmable gate array (FPGA) and its classification performance is tested with publicly available epilepsy dataset.

Spectral information of EEG signals with respect to epilepsy classification

“…In the next stage, a four-level decomposition was applied and the calculated autoregressive (AR) parameters of each sub-band were fed to an MLP classifier. Wang et al [14] presented a novel classification algorithm based on a voting strategy and a hardware implementation. The authors used a band-pass filter to focus only to the 0-32-Hz range and then applied a three-level decomposition and extracted the sample entropy (SampEn) only by the detail coefficients (D1, D2, D3).…”

“…Most of the studies for epileptic activity detection/classification using EEG signal processing, formulate methodologies that analyse the EEG signal by extracting informative features from it [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20]. To this end, spectral analysis of the EEG signal is essential, since epileptic activity interrupts normal brain functionality.…”

“…The EEG spectral analysis is based on a set of frequency sub-bands. Researchers have mainly used wavelet transform (WT) [3][4][5][6][7][8][9][10][11][12][13][14][15][16] and time-frequency distributions (TFD) [17][18][19][20] to analyse the EEG spectral patterns. However, although spectral analysis is a well-known approach, with numerus studies including spectral characteristics in the features extracted from the EEG, the importance of the frequency sub-bands that are used to analyse the signal has never been thoroughly investigated in the literature.…”

“…However, although spectral analysis is a well-known approach, with numerus studies including spectral characteristics in the features extracted from the EEG, the importance of the frequency sub-bands that are used to analyse the signal has never been thoroughly investigated in the literature. It is medically established that brainwaves are divided based on their frequency into several sub-bands, being delta (1-4 Hz), theta (4)(5)(6)(7)(8), alpha (8)(9)(10)(11)(12)(13), beta (13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30) and gamma [21]. Thus, several researchers roughly focus on these sub-bands [3-14, 17, 18], with the technical limitations that the analysis technique imposes (i.e.…”

“…WT). Thus, the importance of the frequency sub-bands and their limits have not been analysed in the literature, since in WT-based approaches the frequency sub-bands are automatically set [3][4][5][6][7][8][9][10][11][12][13][14][15][16], while in TFD-based methodologies, an attempt to compare the impact of different sub-bands has been presented [17], however not being a systematic approach since only four different sub-band combinations were analysed.…”