Assessment of dyslexic children with EOG signals: Determining retrieving words/re-reading and skipping lines using convolutional neural networks. (April 2021)

• Record Type: Journal Article
• Authors: Latifoğlu, Fatma; İleri, Ramis; Demirci, Esra
• Is Part Of: Chaos, solitons and fractals. Volume 145(2021)
• Publication Date: 2021-04
• Subjects: Electrooculography -- EOG -- 2 Dimensional Convolutional Neural Network -- 2D-CNN -- Eye movement -- Dyslexia; Chaotic behavior in systems -- Periodicals; Solitons -- Periodicals; Fractals -- Periodicals; Chaotic behavior in systems; Fractals; Solitons; Periodicals; 003.7
• DOI: 10.1016/j.chaos.2021.110721
• Languages: English
• ISSNs: 0960-0779

Abstract

Highlight: EOG signals were recorded during the reading of a text from healthy and from dyslexic children. To assist in the diagnosis and follow-up of dyslexia is proposed by determining skipping lines and back to eye movement (retrieving words/re-reading movements). Using the proposed method, skipping lines while reading and back to eye movement (retrieving words/re-reading) movements are determined from EOG signals Spectrogram images of movement signals were obtained. These spectrogram images were classified using the 2 dimensional Convolutional Neural Network (2D-CNN) classifier. The 2D-CNN model has classified the skipping lines signal while reading and back to eye movement (retrieving words/re-reading) signal with 99% success. Abstract: This study aims to determine and classify the back to eye movement (retrieving words/re-reading) and skipping lines while reading from electrooculography (EOG) signals. For this aim, EOG signals were recorded during the reading of a text from healthy and from dyslexic children. In this study, a method to assist in the diagnosis and follow-up of dyslexia is proposed by determining skipping lines and back to eye movement (retrieving words/re-reading) while reading. Using the proposed method, skipping lines while reading and back to eye movement (retrieving words/re-reading movements) were determined from EOG signals and spectrogram images of these movement signals are obtained using the Short Time Fourier Transform (STFT) method. These spectrogram images were classified using the 2 Dimensional Convolutional Neural Network (2D-CNN) classifier. The 2D-CNN model has classified the skipping lines signals while reading and back to eye movement (retrieving words/re-reading) signals with 99% success. The findings show that the method proposed in the diagnosis and follow-up of dyslexia can give positive results using these EOG signals.