Personalized Functional Connectivity Based Spatio-Temporal Aggregated Attention Network for MCI Identification.

Area of Science:

• Neuroimaging
• Artificial Intelligence
• Biomedical Engineering

Background:

• Resting-state functional connectivity (rs-fMRI) networks are key biomarkers for mild cognitive impairment (MCI).
• Existing methods often overlook inter-subject functional variations and underutilize temporal fMRI features.
• Current approaches typically rely on group-averaged brain templates, limiting personalized diagnostic capabilities.

Purpose of the Study:

• To develop a novel personalized functional connectivity based dual-branch graph neural network with spatio-temporal aggregated attention (PFC-DBGNN-STAA) for enhanced MCI identification.
• To address limitations in existing methods by incorporating individualized functional variations and temporal dynamics.
• To improve the accuracy and robustness of MCI detection using advanced machine learning techniques.

Main Methods:

• Constructing a personalized functional connectivity (PFC) template to align functional regions and extract individualized features.
• Employing a dual-branch graph neural network (DBGNN) to aggregate individual and group-level template features, considering cross-template dependencies.
• Utilizing a spatio-temporal aggregated attention (STAA) module to capture complex spatial and dynamic relationships within fMRI data.

Main Results:

• The PFC-DBGNN-STAA method achieved high classification accuracies: 90.1% (NC vs. EMCI), 90.3% (EMCI vs. LMCI), and 83.3% (NC vs. EMCI vs. LMCI).
• The proposed method effectively leverages personalized functional connectivity and spatio-temporal features for improved diagnostic performance.
• Evaluation on 442 samples from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database demonstrated superior performance compared to state-of-the-art methods.

Conclusions:

• The proposed PFC-DBGNN-STAA method significantly enhances mild cognitive impairment identification by integrating personalized functional connectivity and spatio-temporal attention.
• The approach effectively captures inter-subject variability and dynamic brain activity, outperforming existing methods.
• This novel framework holds promise for more accurate and individualized diagnosis of MCI.