Diagnosis of multiple sclerosis from EEG signals using nonlinear methods.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Electroencephalogram (EEG) signals contain crucial information about brain activity and neurological diseases.
• Multiple Sclerosis (MS) is a debilitating neurological condition affecting the central nervous system.
• Accurate and early diagnosis of MS is essential for effective management and treatment.

Purpose of the Study:

• To classify individuals with Multiple Sclerosis (MS) from healthy controls using nonlinear features extracted from EEG signals.
• To evaluate the efficacy of different cognitive tasks in differentiating between MS patients and healthy volunteers.
• To identify optimal feature extraction and classification methods for MS detection from EEG data.

Main Methods:

• EEG signals were recorded from healthy volunteers and MS patients during two distinct attentional tasks: color-luminance detection and direction-of-motion detection.
• Nonlinear features were extracted from both original EEG signals and their sub-bands using time-delay embedding for state-space reconstruction.
• Feature dimensionality reduction was performed using T-test and Bhattacharyya criteria, followed by classification using Support Vector Machines (SVM) and K-Nearest Neighbors (KNN).

Main Results:

• The combination of T-test criterion and SVM classifier yielded the highest classification performance for both tasks.
• Maximum classification accuracies achieved were 93.08% for the direction-based task and 79.79% for the color-luminance-based task.
• Nonlinear dynamic features extracted from EEG signals demonstrated significant effectiveness in discriminating between MS patients and healthy individuals.

Conclusions:

• Nonlinear dynamic features derived from EEG signals are valuable for the diagnosis of Multiple Sclerosis.
• The proposed method, utilizing nonlinear EEG analysis during cognitive tasks, offers a promising non-invasive approach for MS detection.
• Further research into advanced feature extraction and classification techniques could enhance diagnostic accuracy for MS.