Related Experiment Video
Updated: Jul 23, 2025
14:27
Identification of Disease-related Spatial Covariance Patterns using Neuroimaging Data
Published on: June 26, 2013
15.7K
Constructing the Schizophrenia Recognition Method Employing GLCM Features from Multiple Brain Regions and Machine
Şerife Gengeç Benli1, Merve Andaç1
1Department of Biomedical Engineering, Faculty of Engineering, Erciyes University, Kayseri 38280, Turkey.
Diagnostics (Basel, Switzerland)
|July 14, 2023
Summary
Magnetic resonance imaging reveals distinct textural characteristics in specific brain regions, aiding in schizophrenia diagnosis. The left hemisphere
Area of Science:
- Neuroimaging
- Psychiatric Disorders
- Biomarker Discovery
Background:
- Schizophrenia diagnosis is complex and crucial for effective treatment.
- Magnetic resonance (MR) imaging offers potential biomarkers for schizophrenia.
- Analyzing textural characteristics in specific brain regions may differentiate individuals with schizophrenia from healthy controls.
Purpose of the Study:
- To numerically analyze textural differences in bilateral amygdala, caudate, pallidum, putamen, and thalamus regions between schizophrenia patients and healthy controls using structural MR images.
- To identify which brain hemisphere exhibits more distinctive textural features.
- To compare the classification performance of various machine learning methods for schizophrenia detection.
Main Methods:
- Gray Level Co-occurrence Matrix (GLCM) features were extracted from five specific brain regions (amygdala, caudate, pallidum, putamen, thalamus) in both hemispheres.
- Machine learning algorithms including Adaboost, Gradient Boost, eXtreme Gradient Boosting, Random Forest, k-Nearest Neighbors, Linear Discriminant Analysis (LDA), and Naive Bayes were employed for classification.
- Classification success was evaluated based on accuracy, sensitivity, specificity, and Area Under the Curve (AUC).
Main Results:
- Textural features from the five specified brain regions, particularly in the left hemisphere, showed higher classification performance for schizophrenia compared to healthy individuals.
- The Linear Discriminant Analysis (LDA) algorithm achieved superior classification results.
- LDA demonstrated 100% AUC, 94.4% accuracy, 92.31% sensitivity, 100% specificity, and a 91.9% F1 score.
Conclusions:
- Textural characteristics of specific brain regions, rather than the entire brain, are significant indicators for identifying schizophrenia.
- The left hemisphere's textural features show particular promise as biomarkers for schizophrenia detection.
- Machine learning, especially LDA, can effectively utilize these textural features for accurate schizophrenia diagnosis.
