Classification of EEG signals from young adults with dyslexia combining a Brain Computer Interface device and an Interactive Linguistic Software Tool

• Published in: Biomedical signal processing and control Vol. 76; p. 103646
• Main Authors: Christodoulides, Pavlos, Miltiadous, Andreas, Tzimourta, Katerina D., Peschos, Dimitrios, Ntritsos, Georgios, Zakopoulou, Victoria, Giannakeas, Nikolaos, Astrakas, Loukas G., Tsipouras, Markos G., Tsamis, Konstantinos I., Glavas, Euripidis, Tzallas, Alexandros T.
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.07.2022
• Subjects: Brain Computer Interface; Dyslexia; EEG; Learning disorders; Random Forests; Wearable device; Dyslexia; Random Forests; Wearable device; Learning disorders; Brain Computer Interface; EEG
• ISSN: 1746-8094; 1746-8108
• DOI: 10.1016/j.bspc.2022.103646

The magnocellular pathway deficit theory has long been considered to be a possible cause for dyslexia, providing an alternative method to explain auditory and visual processing deficits. Several studies have attempted to classify these deficits with the application of machine learning in anatomical brain imaging, rendering the classification techniques using EEG graph measures both robust and reliable. In this paper, a classification of university students with and without dyslexia is attempted with the use of a Brain Computer Interface (BCI) Device and an Interactive Linguistic Software Tool in order to validate the application of such a device in classifying dyslexia in a higher education population. EEG signals acquired from a wearable, sensory EEG recording device from 12 university students with dyslexia along with 14 typically developed, age matched individuals are recorded, while participants were examined in three different experimental conditions: a) auditory discrimination, b) visual recognition c) visual recognition with background music. Spectral features extracted from each EEG rhythm (δ, θ, α, β, γ) are used to train a Random Forests classifier, aiming to identify quantitative EEG features that characterize dyslexia in different brain regions. Results show high levels of accuracy, sensitivity and specificity (above 95%) in the entire brain, followed by the left and right hemisphere, with the highest discrimination performance reported during the third experimental condition with the presence of background music. Different experimental conditions provide high classification accuracy that results in correct discrimination between higher education students with and without dyslexia.