Investigating Brain Lobe Biomarkers to Enhance Dementia Detection Using EEG Data

Abstract Dementia is a growing global health concern that significantly impacts the quality of life for millions of individuals and imposes substantial burdens on healthcare systems. Early detection and accurate diagnosis are crucial for effective dementia management. Electroencephalography (EEG) has emerged as a non-invasive tool for identifying dementia-related abnormalities and assessing brain function. However, existing EEG-based methods often fail to pinpoint specific biomarkers, particularly brain lobe changes. Brain lobe analysis in EEG is essential for advancing dementia detection and improving diagnostic accuracy. This study aims to address this gap by exploring key brain lobes involved in dementia detection and classification, focusing on Alzheimer’s disease (AD) and Frontotemporal dementia (FTD). We introduce a Short-Time Fourier Transform to generate spectrogram images from EEG signals combined with Convolutional Neural Networks to identify the most critical brain lobes for enhanced dementia detection. We have applied Grad-CAM method to improve result interpretability and offer meaningful insights to the research community. Our experiments on OpenNeuro ds004504 EEG dataset for AD and FTD indicate that the parietal lobe exhibits the most significant changes in both conditions, achieving 95.72% accuracy for FTD and 92.25% for AD, followed by the temporal and frontal lobes. When applying the proposed framework to the entire brain region, we achieved 95.59% accuracy for AD and 93.14% for FTD. The findings from EEG-based brain lobe analysis aid experts in improving diagnostic and monitoring tools for neurodegenerative disorders, thereby advancing the understanding and clinical management of dementias like AD and FTD.