An EEG Classification-Based Method for Single-Trial N170 Latency Detection and Estimation

Abstract

Event-related potentials (ERPs) can reflect the high-level thinking activities of the brain. In ERP analysis, the superposition and averaging method is often used to estimate ERPs. However, the single-trial ERP estimation can provide researchers with more information on cognitive activities. In recent years, more and more researchers try to find an effective method to extract single-trial ERPs, because most of the existing methods have poor generalization ability or suffer from strong assumptions about the characteristics of ERPs, resulting in unsatisfactory results under the condition of a very low signal-to-noise ratio. In this paper, an EEG classification-based method for single-trial ERP detection and estimation was proposed. This study used a linear generated EEG model containing templates of ERP local descriptors which include amplitude and latency, and this model can avoid the invalid assumption about ERPs taken by other methods. The purpose of this method is not to recover the whole ERP waveform but to model the amplitude and latency of ERP components. This method afterwards examined the three machine learning models including logistic regression, neural network, and support vector machine in the EEG signal classification for ERP detection and selected the best performed MLPNN model for detection. To get the utmost out of information produced in the classification process, this study also used extra information to propose a new optimization model, with which outperformed detection results were obtained. Performance of the proposed method is evaluated on simulated N170 and real P50 data sets, and the results show that the model is more effective than the Woody filter and the SingleTrialEM algorithm. These results are also consistent with the conclusion of sensory gating, which demonstrated good generalization ability.

Computational Methods for Physiological Signal Processing and Data Analysis

“…The frequency variants with respect to different time spans can be characterized by the continuous wavelet transform, short-time Fourier transform, and wavelet packet transform [ 13 ]. The research work of Zang et al [ 8 ] has demonstrated the merits of joint time-frequency features in the epileptic seizure detection application.…”

“…Detection of amplitude and latency components in a single-trial EEG signal requires a series of artifact removal, feature extraction, and event-related potential (ERP) identification procedures. In [ 8 ], Zang et al developed an effective machine learning technique for latency detection in single-trial EEG signals, instead of the conventional superposition and average method. Two different EEG data sets (i.e., simulated N170 and real P50 recordings) were tested for the performance evaluation purpose.…”

“…In order to improve the EEG signal quality, Zang et al [ 8 ] implemented the signal preprocessing procedures to remove artifacts. The low-frequency noise was first eliminated using a high-pass finite impulse response (FIR) filter (with the cutoff frequency at 1 Hz).…”

“…The experimental results showed that the multilayer perceptrons was able to provide an accuracy up to 90.69% of N170 latencies on the simulated data set. The method proposed by Zang et al [ 8 ] significantly outperformed the Woody filter in different signal-to-noise (SNR) conditions. The experimental results on the real ERP data set indicated that the P50 amplitudes evoked by the first sound were significantly larger than those evoked by the second sound in three auditory tasks.…”

“…The results evaluated with 5-fold, 10-fold, and 20fold cross-validation methods demonstrated that the hybrid time-frequency features and deep learning neural networks can improve the accuracy performance in the epileptic seizure detection [7]. In order to improve the EEG signal quality, Zang et al [8] implemented the signal preprocessing procedures to remove artifacts. The low-frequency noise was first eliminated using a high-pass finite impulse response (FIR) filter (with the cutoff frequency at 1 Hz).…”