OATAO - Open Archive Toulouse Archive Ouverte Open Access Week

Efficient mental workload estimation using task-independent EEG features

Roy, Raphaëlle N. and Charbonnier, Sylvie and Campagne, Aurélie and Bonnet, Stéphane Efficient mental workload estimation using task-independent EEG features. (2016) Journal of Neural Engineering, 13 (2). 1-10. ISSN 1741-2560

[img]
Preview
(Document in English)

PDF (Author's version) - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
1MB

Official URL: http://dx.doi.org/10.1088/1741-2560/13/2/026019

Abstract

Objective. Mental workload is frequently estimated by EEG-based mental state monitoring systems. Usually, these systems use spectral markers and event-related potentials ( ERPs ) . To our knowledge, no study has directly compared their performance for mental workload assessment, nor evaluated the stability in time of these markers and of the performance of the associated mental workload estimators. This study proposes a comparison of two processing chains, one based on the power in fi ve frequency bands, and one based on ERPs, both including a spatial filtering step ( respectively CSP and CCA ) , an FLDA classification and a 10-fold cross-validation. Approach. To get closer to a real life implementation, spectral markers were extracted from a short window ( i.e. towards reactive systems ) that did not include any motor activity and the analyzed ERPs were elicited by a task-independent probe that required a re fl ex-like answer ( i.e. close to the ones required by dead man ’ s vigilance devices ) . The data were acquired from 20 participants who performed a Sternberg memory task for 90 min ( i.e. 2 / 6 digits to memorize) inside which a simple detection task was inserted. The results were compared both when the testing was performed at the beginning and end of the session. Main results. Both chains performed significantly better than random; however the one based on the spectral markers had a low performance ( 60% ) and was not stable in time. Conversely, the ERP-based chain gave very high results ( 91% ) and was stable in time. Significance. This study demonstrates that an efficient and stable in time workload estimation can be achieved using task-independent spatially filtered ERPs elicited in a minimally intrusive manner.

Item Type:Article
Audience (journal):International peer-reviewed journal
Uncontrolled Keywords:
Institution:French research institutions > Commissariat à l'Energie Atomique et aux énergies alternatives - CEA (FRANCE)
French research institutions > Centre National de la Recherche Scientifique - CNRS (FRANCE)
Other partners > Université Grenoble Alpes - UGA (FRANCE)
Other partners > Université de Savoie Mont Blanc - USMB (FRANCE)
Laboratory name:
Statistics:download
Deposited By: Raphaëlle ROY
Deposited On:21 Mar 2018 09:44

Repository Staff Only: item control page