Multi Modal Feature Extraction for Classification of Vascular Dementia in Post-Stroke Patients Based on EEG Signal

Area of Science:

• Neuroscience
• Medical Imaging Analysis
• Signal Processing

Background:

• Dementia, particularly vascular dementia, significantly impacts post-stroke patients' cognitive functions.
• Current diagnostic methods, including brain imaging, are costly and time-consuming.
• Traditional electroencephalogram (EEG) analysis is subjective and requires expert interpretation.

Purpose of the Study:

• To investigate the efficacy of quantitative EEG (QEEG) analysis for diagnosing vascular dementia in post-stroke individuals.
• To develop a more accessible and objective diagnostic tool for cognitive impairment after stroke.

Main Methods:

• Utilized 19-channel EEG recordings from three groups: healthy elderly, post-stroke with mild cognitive impairment, and post-stroke with dementia.
• Extracted features using QEEG, including relative power, coherence, and signal complexity.
• Employed Support Vector Machine (SVM) and K-Nearest Neighbor (KNN) algorithms for classification.

Main Results:

• Achieved a highest classification accuracy of 96% using Gaussian SVM.
• Demonstrated high sensitivity (95.6%) and specificity (97.9%) in distinguishing between cognitive states.
• QEEG analysis effectively differentiated between normal cognition, mild cognitive impairment, and dementia in post-stroke patients.

Conclusions:

• QEEG analysis presents a promising, cost-effective alternative for diagnosing vascular dementia in post-stroke patients.
• This method can serve as an additional diagnostic criterion, improving patient management and quality of life.
• Further integration of QEEG could enhance early detection and intervention strategies for cognitive decline post-stroke.