Two-stage Hardware-Friendly Epileptic Seizure Detection Method with a Dynamic Feature Selection

Abstract

A novel low-complexity method of detecting epileptic seizures from intracranial encephalography (iEEG) signals is presented. In the proposed algorithm, coastline, energy and nonlinear energy features of iEEG signals are extracted in a patient-specific two-stage seizure detection system. The detection stage of the proposed system, which extracts two times more features than the monitoring stage, is only powered on when the monitoring stage detects a seizure occurrence. A new metric is defined to demonstrate the significance of the two-stage architecture and show the time duration over which the detection stage is activated. The new parameter is called detection stage activation ratio (DAR) and it is equal to 0.272 in this work. In addition, the proposed seizure detector outperforms other algorithms which utilize a single feature or multiple features continuously in terms of sensitivity, specificity and DAR. Therefore, it is highly suitable for seizure detector implants in which reducing the power consumption is a critical factor to increase the lifetime of the implanted battery. The algorithm is implemented on a Cyclone V FPGA and has a low dynamic power of 1 𝛍W when tested on human iEEG signals of six patients from the Bern Inselspital dataset. It reaches a perfect sensitivity of 100% tested on 120 hours of iEEG data containing 24 seizure periods of six patients.

Computation Complexity Reduction Technique for Accurate Seizure Detection Implants

“…Table III compares the performance of the proposed design (the average results of patients) with recently published state-of-the-art that also uses the SWEC-ETHZ dataset. The MATLAB simulation results evidence a sensitivity of 100% as [9] which is better than other works. In terms of specificity and detection delay, the presented method outperforms the state-of-the-art.…”

“…Also, the computational complexity of our algorithm reflected by the C.D parameter is outstandingly lower than the other works. Using the effective number of features for two-stage architecture systems and the average number of electrode channels of all patients cause floating point values of C.D for [2], [8] and [9].…”

Hardware-Friendly Random Forest Classification of iEEG Signals for Implantable Seizure Detection

“…Furthermore, the hardware implementation of HD classifier is not performed. [5] extracts Line-length, energy and nonlinear energy time-domain features in a two-stage architecture to provide a low-power feature extraction method. [3] presents patient-specific channel and feature selection approaches based on time-domain features that are highly suitable for accurate and low-complexity seizure detection.…”

“…[3] presents patient-specific channel and feature selection approaches based on time-domain features that are highly suitable for accurate and low-complexity seizure detection. However, [3], [4] and [5] do not perform hardware implementation of a machine learning classifier that is compatible with seizure detector implants.…”