Subnetwork mining on functional connectivity network for classification of minimal hepatic encephalopathy

Area of Science:

• Neuroscience
• Medical Imaging
• Computational Biology

Background:

• Hepatic encephalopathy (HE) is a severe complication of cirrhosis, impacting brain function.
• Minimal HE (MHE) is an intermediate stage requiring accurate diagnosis for timely intervention.
• Existing brain network analyses often overlook local network structures (subnetworks).

Purpose of the Study:

• To develop a novel method for identifying disease-related subnetworks in MHE.
• To improve the classification accuracy of MHE using brain network analysis.
• To gain insights into the pathological basis of MHE through subnetwork identification.

Main Methods:

• Proposed a frequent-subnetwork-based approach for MHE classification.
• Mined frequent subnetworks separately for MHE and non-HE (NHE) patient groups.
• Employed a graph-kernel method to select discriminative subnetworks for classification.

Main Results:

• The proposed method achieved improved classification performance compared to existing network-based techniques.
• Successfully identified disease-related subnetworks associated with MHE.
• The identified subnetworks offer potential for better understanding MHE pathology.

Conclusions:

• The novel frequent-subnetwork mining method enhances MHE classification accuracy.
• This approach effectively identifies key brain network alterations in MHE.
• The findings contribute to a deeper understanding of MHE's neurological underpinnings.