Network Optimization of Functional Connectivity Within Default Mode Network Regions to Detect Cognitive Decline.

Area of Science:

• Neuroscience
• Medical Imaging
• Network Analysis

Background:

• Global population aging increases cognitive decline and degenerative brain diseases.
• Mild cognitive impairment diagnoses are clinical; functional magnetic resonance imaging (fMRI) shows default mode network (DMN) connectivity decline correlates with neurological disorders, especially prodromal Alzheimer's disease.

Purpose of the Study:

• Develop a novel network analysis technique using fMRI data to characterize transition stages from healthy brain aging to cognitive decline.
• Focus on intra-nodal DMN connectivity, unlike previous inter-nodal focus, by incorporating sparsity into the k-cardinality tree (KCT) problem.

Main Methods:

• Developed new mathematical formulations for the NP-hard KCT problem.
• Implemented a fast heuristic approach to efficiently solve KCT models for large DMN regions.
• Applied the technique to fMRI data to identify sparse connectivity patterns within DMN regions.

Main Results:

• Traditional fMRI group analysis failed to detect significant DMN connectivity differences between normal aging and cognitively impaired subjects.
• The proposed KCT approaches demonstrated higher sensitivity than the regional homogeneity approach.
• Significant differences were detected in left and right medial temporal regions of the DMN.

Conclusions:

• The novel KCT-based network analysis is more sensitive in detecting subtle DMN connectivity changes associated with cognitive decline.
• This technique offers a promising tool for characterizing early-stage Alzheimer's disease and related conditions.
• Highlights the importance of intra-nodal connectivity and sparsity in understanding brain network dynamics.