Detection of Negative Stress through Spectral Features of Electroencephalographic Recordings and a Convolutional Neural Network

Abstract

In recent years, electroencephalographic (EEG) signals have been intensively used in the area of emotion recognition, partcularly in distress identification due to its negative impact on physical and mental health. Traditionally, brain activity has been studied from a frequency perspective by computing the power spectral density of the EEG recordings and extracting features from different frequency sub-bands. However, these features are often individually extracted from single EEG channels, such that each brain region is separately evaluated, even when it has been corroborated that mental processes are based on the coordination of different brain areas working simultaneously. To take advantage of the brain’s behaviour as a synchronized network, in the present work, 2-D and 3-D spectral images constructed from common 32 channel EEG signals are evaluated for the first time to discern between emotional states of calm and distress using a well-known deep-learning algorithm, such as AlexNet. The obtained results revealed a significant improvement in the classification performance regarding previous works, reaching an accuracy about 84%. Moreover, no significant differences between the results provided by the diverse approaches considered to reconstruct 2-D and 3-D spectral maps from the original location of the EEG channels over the scalp were noticed, thus suggesting that these kinds of images preserve original spatial brain information.

Enhancing EEG-Based Mental Stress State Recognition using an Improved Hybrid Feature Selection Algorithm

“…Various time-domain, frequency-domain, and time-frequency-domain features have been proposed in previous studies for mental stress detection [13,15,16]. However, little research utilizes connectivity features.…”

“…To begin, we conducted our study using a fixed time window length; however, experimenting with alternative window lengths can aid in model generalization. Additionally, It is worth exploring, data augmentation via sliding windows as this technique is increasingly being utilized in combination with deep learning on EEG [61]. Second, a more in-depth examination of other connective network features should be explored since they give detailed information about channel interconnections.…”

“…In multi-channel EEG, several features from the time domain, frequency domain, time-frequency domain, spatial domain, etc., contribute to the high dimensional feature space in which one aims to recognize or assess several brain states such as seizure detection (epilepsy) [9], motor imaginary [10], depression [11], emotion detection [12,13], and mental stress recognition [14]. Recently, EEG signals have been used extensively in the field of emotion recognition, particularly in the recognition of distress due to its harmful influence on physical and mental health [15,16]. However, one of the major challenges in building a successful model for stress detection is finding the most appropriate features.…”

Epileptic Seizures Detection Using Deep Learning Techniques: A Review

“…Nowadays, DL models have made considerable advancements [ 160 , 161 , 162 , 163 , 164 ]. This has resulted in the development of computer hardware [ 165 , 166 ] that is expensive and not easily accessible to the researchers.…”

Enhancing EEG-Based Mental Stress State Recognition Using an Improved Hybrid Feature Selection Algorithm

“…Recently, EEG signals have been used extensively in the field of emotion recognition, particularly in recognition of distress due to harmful influences on physical and mental health [1,23]. However, one of the significant challenges in building a successful model for stress detection is finding the most appropriate features.…”