An SVM-based system and its performance for detection of seizures in neonates

Abstract

This work presents a multi-channel patient-independent neonatal seizure detection system based on the SVM classifier. Several post-processing steps are proposed to increase temporal precision and robustness of the system and their influence on performance is shown. The SVM-based system is evaluated on a large clinical dataset using several epoch-based and event based metrics and curves of performance are reported. Additionally, a new metric to measure the average duration of a false detection is proposed to accompany the event-based metrics.

Improved multi-stage neonatal seizure detection using a heuristic classifier and a data-driven post-processor

“…The following methods have been considered: time-frequency based analysis and multi-layer perceptrons (MLPs) (Hassanpour et al 2004), quantitative features and a linear discriminant classifier (Greene et al 2008), support vector machine (SVM) based classifier (Temko et al 2009), Bayesian classifier via Gaussian mixture models (Temko et al, 2009), adaptive multi-channel information fusion (Li and Jeremic 2011), SVM classifier and Kalman filter (Bogaarts et al 2014), and trend template analysis with SVM classifier tested on fetal lambs (Zwanenburg et al 2015).…”

Dynamic, location-based channel selection for power consumption reduction in EEG analysis

“…The decisions from all EEG channels are smoothed and fused in order to produce a final decision of seizure or non-seizure. The performance of this system has been previously published and discussed in [11], with improvements in performance continued in [12] and [13]. As shown in this section, the REACT system is a complex algorithm and is computationally intensive.…”

Speech recognition features for EEG signal description in detection of neonatal seizures

“…Recent examples of this methodology involve the work in [1][2] [3]. The second approach relies on classifier based methods which employ elements of pattern recognition to classify a set of features using a data-driven decision rule [4][5] [6].…”