Brain Morphometry Methods for Feature Extraction in Random Subspace Ensemble Neural Network Classification of...

Area of Science:

• Neuroimaging
• Machine Learning
• Psychiatry

Background:

• Schizophrenia diagnosis can be challenging.
• Machine learning (ML) offers potential for developing auxiliary diagnostic tools.
• Neuroimaging data, particularly magnetic resonance imaging (MRI), provides valuable structural information.

Purpose of the Study:

• To present a novel classification framework for schizophrenia diagnostics.
• To evaluate the effectiveness of combining different feature extraction methods from MRI data.
• To assess the performance of a multilayer perceptron (MLP) classifier within this framework.

Main Methods:

• Features were extracted from MRI data using voxel-based morphometry (VBM) and deformation-based morphometry (DBM).
• A random subspace ensemble-based artificial neural network, specifically a multilayer perceptron (MLP), was employed as the classifier.
• The framework was tested on data from first-episode schizophrenia patients and healthy controls.

Main Results:

• Combining VBM and DBM features enhanced MLP classification accuracy to 73.12%.
• This represents a 5% improvement compared to using VBM or DBM features alone.
• Comparisons with support vector machines (SVMs) did not reveal a superior classifier.

Conclusions:

• The integration of diverse morphometric features from MRI significantly improves ML-based schizophrenia classification.
• The proposed framework demonstrates the potential of ML as an auxiliary diagnostic tool in psychiatry.
• Further research is needed to determine the optimal classifier for this application.