Voxel-based morphometry analysis and machine learning based classification in pediatric mesial temporal lobe epilepsy...

Area of Science:

• Neuroimaging
• Machine Learning
• Epilepsy Research

Background:

• Mesial temporal lobe epilepsy with hippocampal sclerosis (MTLE-HS) is a frequent pediatric epilepsy syndrome.
• Accurate classification of MTLE-HS is crucial for effective treatment and management.
• Current diagnostic methods may benefit from advanced computational techniques.

Purpose of the Study:

• To assess the feasibility of combining Voxel-Based Morphometry (VBM) and Support Vector Machine (SVM) for classifying pediatric MTLE-HS.
• To evaluate the diagnostic accuracy of this combined neuroimaging and machine learning approach.
• To correlate VBM findings with clinical observations in MTLE-HS.

Main Methods:

• Acquisition of 3D T1-weighted MRI scans from 37 participants (22 with MTLE-HS, 15 healthy controls).
• Application of VBM to identify gray matter volume abnormalities in MTLE-HS patients.
• Utilizing SVM with leave-one-out cross-validation to classify MTLE-HS based on VBM-derived gray matter volumes.

Main Results:

• VBM analysis revealed both gray matter volume increases and reductions in participants with MTLE-HS compared to controls.
• SVM classification achieved high accuracy, with Area Under the Curve (AUC) values of 0.870 (Left HS vs. HC), 0.976 (Right HS vs. HC), and 0.902 (HS vs. HC).
• VBM findings demonstrated concordance with established clinical characteristics of MTLE-HS.

Conclusions:

• The combination of VBM and SVM is a feasible and highly accurate method for classifying pediatric MTLE-HS.
• This integrated approach offers a promising tool for the objective diagnosis of MTLE-HS.
• The findings support the clinical utility of advanced neuroimaging analysis in pediatric epilepsy.