Electroencephalogram-based decoding cognitive states using convolutional neural network and likelihood ratio based score fusion.

Raheel Zafar,Sarat C Dass,Aamir Saeed Malik,Bin He

doi:10.1371/journal.pone.0178410

Raheel Zafar, Sarat C Dass + Show 2 more

Open Access

https://doi.org/10.1371/journal.pone.0178410

Copy DOI

Journal: PloS one	Publication Date: May 30, 2017
Citations: 32	License type: CC BY 4.0

Affiliation: Universiti Teknologi Petronas

Abstract

Electroencephalogram (EEG)-based decoding human brain activity is challenging, owing to the low spatial resolution of EEG. However, EEG is an important technique, especially for brain–computer interface applications. In this study, a novel algorithm is proposed to decode brain activity associated with different types of images. In this hybrid algorithm, convolutional neural network is modified for the extraction of features, a t-test is used for the selection of significant features and likelihood ratio-based score fusion is used for the prediction of brain activity. The proposed algorithm takes input data from multichannel EEG time-series, which is also known as multivariate pattern analysis. Comprehensive analysis was conducted using data from 30 participants. The results from the proposed method are compared with current recognized feature extraction and classification/prediction techniques. The wavelet transform-support vector machine method is the most popular currently used feature extraction and prediction method. This method showed an accuracy of 65.7%. However, the proposed method predicts the novel data with improved accuracy of 79.9%. In conclusion, the proposed algorithm outperformed the current feature extraction and prediction method.

Highlights

Decoding brain activity involves the reconstruction of stimuli or brain state from the information measured using different modalities like electroencephalogram (EEG) or functional magnetic resonance imaging
For decoding the brain activity patterns, multivariate pattern analysis (MVPA) is an emerging technique and has proven as a highly useful technique for decoding of different patterns of brain activity [11, 53, 54]. This new concept of MVPA and more EEG channel device encourages neuroscientists to decode brain activity using EEG, as EEG with a higher number of channels, improves spatial resolution. This is the reason we proposed an algorithm to improve the decoding accuracy with EEG data which has new machine learning technique (ConvNet) and a different prediction method likelihood ratio based score fusion (LRBSF) in neuroscience
It consists of comprehensive data arrangement, a modification of the Convolutional neural network (ConvNet) model, a t-test and LRBSF, which predicts the novel data of different object categories with an average prediction accuracy of 79.9%

Summary

Introduction

Decoding brain activity involves the reconstruction of stimuli or brain state from the information measured using different modalities like electroencephalogram (EEG) or functional magnetic resonance imaging (fMRI). Stimulus or brain state information is encoded in the brain and is present in the form of neuronal activity. Decoding this recorded neural information and associated changes in brain activity can be used to predict the specific tasks or stimuli that caused the response in the brain. Different neuroimaging techniques can be used to find differences in brain activity during different tasks or conditions. In fMRI, the measured signal is the blood-oxygen-level dependent signal. EEG records electrical signals that indicate activity and responses in the brain

Methods

Results

Discussion

Conclusion