Novel Effective Connectivity Inference Using Ultra-Group Constrained Orthogonal Forward Regression and Elastic...

Area of Science:

• Neuroscience
• Medical Imaging
• Machine Learning

Background:

• Mild cognitive impairment (MCI) detection is vital for preventing progression to Alzheimer's disease (AD).
• Functional magnetic resonance imaging (fMRI) derived brain connectivity networks are used for MCI/AD identification.
• Understanding effective connectivity is essential for diagnosing MCI.

Purpose of the Study:

• To propose a novel sparse constrained effective connectivity inference method for MCI identification.
• To develop an elastic multilayer perceptron classifier for improved MCI detection.
• To identify neuroimaging biomarkers associated with MCI.

Main Methods:

• Designed an ultra-group constrained structure detection algorithm for effective connectivity network topology.
• Employed an ultra-orthogonal forward regression algorithm to construct the effective connectivity network.
• Utilized an elastic multilayer perceptron classifier for MCI identification based on the constructed network.

Main Results:

• Achieved high classification accuracy for MCI identification compared to state-of-the-art methods.
• Identified a loss of rich club effect in MCI patients.
• Observed decreased connectivity among specific brain regions in individuals with MCI.

Conclusions:

• The proposed method enhances MCI classification performance.
• The study successfully discovered critical disease-related neuroimaging biomarkers for MCI.
• Findings contribute to understanding brain network alterations in MCI.