Periodogram Connectivity of EEG Signals for the Detection of Dyslexia

Abstract

Electroencephalography (EEG) signals provide an important source of information of brain activity at different areas. This information can be used to diagnose brain disorders according to different activation patterns found in controls and patients. This acquisition technology can be also used to explore the neural basis of less evident learning disabilities such as Developmental Dyslexia (DD). DD is a specific difficulty in the acquisition of reading skills not related to mental age or inadequate schooling, whose prevalent is estimated between 5% and 12% of the population. In this paper we propose a method to extract discriminative features from EEG signals based on the relationship among the spectral density at each channel. This relationship is computed by means of different correlation measures, inferring connectivity-like markers that are eventually selected and classified by a linear support vector machine. The experiments performed shown AUC values up to 0.7, demonstrating the applicability of the proposed approach for objective DD diagnosis.

Detecting Phase-Synchrony Connectivity Anomalies in EEG Signals. Application to Dyslexia Diagnosis

“…Dyslexia classification using biomedical signals is also a current research trend, oppositely to the traditional method based only on behavioural tests. Works using structural imaging [67], MEG [68] and EEG [27][28][29] signals are shown in Table 3. As in [68], the classification problem with biomedical signals is usually addressed using different features extracted from MEG signals, obtaining accuracy values up to 0.93.…”

“…Then, temporal and spectral features from the EEGs at different frequency were used to build the feature space, obtaining AUC values ranging from 0.69 to 0.89. On the other hand, [29] uses periodogram-based features. Specifically, the spectral density is obtained from the EEGs and then Principal Component Analysis is used to reduce the dimension of the spectral density vector.…”

Initial Study on Implementation of the Low-Frequency Wave Markers for the Purpose of Diagnostic Tests’ Performance and Neurofeedback Therapy

“…The authors decided to implement the Welch's periodogram method, which is an improved version of the typical (standard) periodogram (Martinez-Murcia et al, 2019;Mostile et al, 2019;Mulkey et al, 2019). The chosen method is the modified periodogram, which is an improved estimator of the Power Spectral Density (PSD) consisting of the time-series divided into segments.…”

“…Classification of biomedical data is a problem due to the nature of such signals, therefore this task is very challenging; however, classification of spectral power features of EEG data is currently quite well documented. Proper data-classification plays a crucial role in diagnostics and later in choosing the appropriate therapy (Kawala-Janik et al, 2018;Lazar et al, 2019;Martinez-Murcia et al, 2019).…”

Artificial intelligence within the interplay between natural and artificial computation: Advances in data science, trends and applications

“…Novel advances related to EEG processing include [136], where correlation between EEG spectrum in the typical EEG bands (Delta, Theta, Alpha, Beta and Gamma), are used to infer functional connectivity patterns that reveals differences between controls and dyslexic subjects. Unlike classical experiments in this area, non-interactive auditory stimuli have been used in this work.…”