VLSI Design of SVM-Based Seizure Detection System With On-Chip Learning Capability

Abstract

Portable automatic seizure detection system is very convenient for epilepsy patients to carry. In order to make the system on-chip trainable with high efficiency and attain high detection accuracy, this paper presents a very large scale integration (VLSI) design based on the nonlinear support vector machine (SVM). The proposed design mainly consists of a feature extraction (FE) module and an SVM module. The FE module performs the three-level Daubechies discrete wavelet transform to fit the physiological bands of the electroencephalogram (EEG) signal and extracts the time-frequency domain features reflecting the nonstationary signal properties. The SVM module integrates the modified sequential minimal optimization algorithm with the table-driven-based Gaussian kernel to enable efficient on-chip learning. The presented design is verified on an Altera Cyclone II field-programmable gate array and tested using the two publicly available EEG datasets. Experiment results show that the designed VLSI system improves the detection accuracy and training efficiency.

An electronic neuromorphic system for real-time detection of high frequency oscillations (HFO) in intracranial EEG

“…As such, this approach can not be directly compared with largescale neuromorphic computing approaches, or state-of-the-art or deep-learning methods. Other embedded systems and VLSI devices designed for the specific case of processing and/or classifying EEG signals have been proposed in recent years [63][64][65][66] . Table 3 highlights the differences between these systems and the one presented in this work.…”

“…However, this design is missing a co-integrated analog headstage and, by extension, an integrated local binary pattern encoder. Separating the signal encoding stage from the processing stage allows the implementation of sophisticated signal processing techniques and machine learning algorithms, as is evident from the works of Burello et al 63 and Feng et al 64 . But using off-the shelf platforms for signal encoding, processing, or both, leads to much higher power consumption and bulky platforms that make the design of compact and portable embedded systems more challenging.…”

Hardware Acceleration of High Sensitivity Power-Aware Epileptic Seizure Detection System Using Dynamic Partial Reconfiguration

“…The only division that cannot be removed, without affecting the sensitivity obtained, is the division by the standard deviation in the Hurst exponent shown in equation (6). The standard deviation is not a constant, but rather varying according to the values of the data samples in each window.…”

“…An implantable device that can be inserted in the patient's scalp providing an electrical stimulation as soon as a seizure occurs can be useful for the patient. Recently, machine learning techniques are exploited in automatic seizure detection algorithms as reported in [6][7][8]. Machine learning is the science of teaching computers how to deal with different situations and to perform some complicated tasks without being programmed.…”

Automated Seizure Detection Tailored Towards Portable Patient Care Using A Novel, Computationally Efficient Method of Analysis

“…Computer-aided detection systems have been explored since the early 1970s (Baldassano et al, 2017;Saini and Dutta, 2017). Lately, many automated detection systems based on machine learning algorithms have been proposed (Feng et al, 2018;Liu et al, 2012;Saini and Dutta, 2017;Zhang and Chen, 2017). A plethora of feature extraction techniques have been implemented, from Fourier and wavelet transformations to non-linear dynamics such as fractal dimensions.…”