OATAO - Open Archive Toulouse Archive Ouverte Open Access Week

Mental Workload Classification with fNIRS using Temporal Convolutional Networks

Rayudu, Venkata Suresh and Gharpurey, Ranjit Mental Workload Classification with fNIRS using Temporal Convolutional Networks. (2020) In: 1st International Conference on Cognitive Aircraft Systems - ICCAS 2020, 18 March 2020 - 19 March 2020 (Toulouse, France). (Unpublished)

[img]
Preview
(Document in English)

PDF - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
3MB

Abstract

Neuroimaging classification with functional Near Infrared Spectroscopy (fNIRS) can be used for applications such as Brain Computer Interface (BCI) and Brain Machine Interface (BMI). BCI/BMI provide a means for decoding brain signals into actions, thus providing a means of communication for people suffering with paralysis such as conditions such as locked-in syndrome (LIS), spinal cord injury [3]. This paper demonstrates that fNIRS can be used effectively for BCI using four-way classification of left and right motor imagery (MI), mental arithmetic (MA) and rest tasks. 36-channel fNIRS data, capturing hemodynamic signals from frontal, motor and visual cortex from 29 subjects during an experimental paradigm consisting of left, right motor imagery, mental arithmetic and rest states is used. The data is obtained from an open access dataset 1. Each experiment in 1 consisted of six sessions: three sessions of left and right-hand MI and three sessions of MA and baseline tasks (taking a rest without any thought). Each session consisted of a 60-s pre- and post-experiment rest period, and 20 repetitions of the task (LMI/RMI/MA/rest). fNIRS optical intensity signals are converted into HbO and HbR concentration changes using modified Beer Lambert Law (mBLL). The fNIRS sampling frequency is 10-Hz. HbO and HbR are applied to a third-order bandpass Chebyshev filter with cut-off frequencies of 0.01-Hz and 0.1-Hz to attenuate physiological noise caused by respiration, motion artifact and heartbeat.

Item Type:Invited Conference
Audience (conference):International conference without published proceedings
Uncontrolled Keywords:
Institution:Other partners > University of Texas at Austin (USA)
Statistics:download
Deposited On:10 May 2021 18:38

Repository Staff Only: item control page