Multimodal MRI Analysis for Schizophrenia Classification Using Machine Learning Algorithm

Schizophrenia is a complicated mental condition marked by disruptions in thought processes, perceptions, and emotional responses, which can lead to severe social and occupational failure. Structural Magnetic Resonance Imaging (sMRI) and resting-state Functional Magnetic Resonance Imaging (rs-fMRI) p...

Full description

Saved in:
Bibliographic Details
Published in2024 International Conference on Emerging Techniques in Computational Intelligence (ICETCI) pp. 451 - 458
Main Authors S, Tamilarasi, Rajangam, Vijayarajan
Format Conference Proceeding
LanguageEnglish
Published IEEE 22.08.2024
Subjects
Accuracy
Brain modeling
Classification algorithms
Energy
Feature extraction
fMRI
Functional magnetic resonance imaging
Heuristic algorithms
Machine learning algorithms
Mean
Mental disorders
schizophrenia
Sensitivity and specificity
sMRI
Support vector machines
SVM
Variance
Online AccessGet full text

Cover

More Information
Summary:Schizophrenia is a complicated mental condition marked by disruptions in thought processes, perceptions, and emotional responses, which can lead to severe social and occupational failure. Structural Magnetic Resonance Imaging (sMRI) and resting-state Functional Magnetic Resonance Imaging (rs-fMRI) provide useful information about the neuroanatomical and functional changes associated with schizophrenia. This paper presents a unique approach to diagnosing schizophrenia by extracting energy, mean, and variance features from both sMRI and rs-fMRI data. Support Vector Machine (SVM) algorithm is used to conduct binary classification on sMRI -derived and rs-fMRI -derived features. sMRI -derived features detect dimensional structural variations in brain anatomy, whereas fMRI -derived features show dynamic functional activity patterns over time. Preliminary results show that both modalities have promising discriminating power, with fMRI -derived features outperforming sMRI-derived features. On a cohort of 106 schizophrenia patients and 160 healthy controls, our proposed technique achieved 93.5 % and 95.6 % classification accuracy for sMRI and rs-fMRI data, respectively. The classification accuracies from both modalities are compared to determine their diagnostic usefulness for schizophrenia. Integrating both modalities has the potential to improve the sensitivity and specificity of diagnostic algorithms for schizophrenia, resulting in better clinical results and customized treatment methods. These findings highlight the significance of multimodal MRI methods in schizophrenia diagnosis,
DOI:10.1109/ICETCI62771.2024.10704116