Multimodal multiview bilinear graph convolutional network for mild cognitive impairment diagnosis

Area of Science:

• Neuroimaging and computational neuroscience
• Biomedical data analysis and machine learning

Background:

• Mild cognitive impairment (MCI) is a critical precursor to Alzheimer's disease (AD), necessitating advanced diagnostic tools.
• Existing diagnostic methods often overlook valuable non-imaging data (e.g., genetic, clinical) and struggle with heterogeneity in imaging data.
• The integration of multimodal data remains a challenge due to data heterogeneity and the need to capture complex relationships.

Purpose of the Study:

• To propose a novel Multimodal Multiview Bilinear Graph Convolution (MMBGCN) framework for improved disease risk prediction.
• To effectively integrate neuroimaging (MRI) and non-imaging data for a more comprehensive disease assessment.
• To address data heterogeneity and noise in network construction for more robust diagnostic models.

Main Methods:

• Extraction of grey matter (GM), white matter (WM), and cerebrospinal fluid (CSF) features from MRI data.
• Construction of a multimodal shared adjacency matrix by combining MRI features with non-imaging data (genetic, clinical).
• Development of a multiview network using the shared adjacency matrix, followed by weighted MRI feature extraction and bilinear convolution for spatial pattern restoration.
• Integration of recovered spatial patterns with genetic information for final disease prediction.

Main Results:

• The proposed MMBGCN framework demonstrated superior performance on the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset.
• The framework achieved an average classification accuracy of 89.6% in binary classification tasks.
• MMBGCN outperformed existing related algorithms in disease risk prediction, highlighting its effectiveness.

Conclusions:

• The MMBGCN framework offers a robust and effective approach for disease risk prediction by leveraging multimodal data.
• The study highlights the importance of integrating diverse data types for accurate diagnosis of conditions like MCI and AD.
• The proposed method facilitates advancements in MCI diagnosis and provides a valuable tool for future research in neurodegenerative diseases.