Multimetric brain functional-structural connectivity mutual coupling for mild cognitive impairment identification.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Biomedical Engineering

Background:

• Mild cognitive impairment (MCI) is a precursor to neurodegenerative diseases like Alzheimer's.
• Multimodal MRI, combining functional connectivity (FC) and structural connectivity (SC), offers insights into brain network organization.
• Existing methods often use single metrics and unidirectional models, potentially missing complex brain network dynamics.

Purpose of the Study:

• To develop an advanced deep learning framework, the Multimetric Mutual Coupling Network (MMCN), for accurate MCI identification.
• To leverage multiple quantitative metrics for both FC and SC to create richer brain network representations.
• To model the intrinsic bidirectional interplay between brain function and structure for improved diagnostic capabilities.

Main Methods:

• Constructing subject-specific multimetric connectivity graphs using diverse FC and SC metrics.
• Implementing an FC/SC-specific patch embedding module for feature extraction.
• Utilizing a cross-guided attention mechanism for bidirectional feature refinement between FC and SC networks.

Main Results:

• The proposed MMCN demonstrated superior performance in MCI identification compared to state-of-the-art methods.
• Experiments on public and local datasets validated the effectiveness of the multimetric mutual coupling approach.
• The study highlighted the importance of bidirectional modeling for capturing subtle connectivity disruptions in MCI.

Conclusions:

• The MMCN provides a robust and accurate method for early identification of mild cognitive impairment.
• Multimetric-based mutual coupling of functional and structural brain connectivity is crucial for enhancing MCI detection.
• The developed framework offers a promising tool for clinical intervention in neurodegenerative disease pathways.