Programme

P.47 -- Posters, VCTP-51

Date: Tuesday, 4 August 2026

Time: 09:30 - 10:30

Frontal-Parietal Disruption in AD and FTD: A Multidimensional EEG Assessment via Symbolic Transfer Entropy

Le Duy Manh1,2, Trinh Xuan Hoang3, Man Minh Tan1,2, Pham Thanh Dam1,2, Dinh Thanh Binh1,2, Nguyen Hoang Long1,4, Ngo Thi Tam1,4, Nguyen Thi Thuy3

1 Institute for Advanced Study in Technology, Ton Duc Thang University, Ho Chi Minh City, 70000, Vietnam 2 Faculty of Applied Sciences, Ton Duc Thang University, Ho Chi Minh City, 70000, Vietnam 3 Institute of Physics, Vietnam Academy of Science and Technology, 10 Dao Tan, Giang Vo, Ha Noi, 100000, Viet Nam 4 Faculty of Pharmacy, Ton Duc Thang University, Ho Chi Minh City, 70000, Vietnam

Tracking the breakdown of large-scale neural networks is vital for identifying cognitive decline, particularly along the frontal-parietal axis. This study evaluates an integrated electroencephalography (EEG) framework designed to map these regional disruptions, aiming to distinguish Alzheimer’s disease (AD) and frontotemporal dementia (FTD) from healthy aging (HC).Resting-state recordings from 88 subjects (36 AD, 23 FTD, 29 HC) were sourced from public repository ds004504. Signals from channels Fz and Pz were modeled using Symbolic Transfer Entropy (Net Flow) alongside complementary spectral-temporal metrics. These parameters were projected into a unified biophysical manifold via Mahalanobis Distance.Statistical testing confirmed distinct group profiles ($p < 0.05$). Net Flow values dropped sequentially from HC to AD, revealing an asymmetrical collapse in directional frontal-parietal signaling. Concurrently, alpha-band slowing and temporal delays escalated, with FTD presenting intermediate degradation. The manifold optimization maximized group boundaries ($p = 2.95 \times 10^{-7}$). An SVM classifier separated HC from AD with 84.6% accuracy (AUC = 0.83). Ultimately, quantifying directional network failure via symbolic information theory serves as an effective, accessible biomarker for dementia classification.

Presenter: Le Duy Manh