Related Experiment Video
Updated: Aug 15, 2025
07:30
Brain Morphology of Cannabis Users With or Without Psychosis: A Pilot MRI Study
Published on: August 18, 2020
6.8K
Framework to Detect Schizophrenia in Brain MRI Slices with Mayfly Algorithm-Selected Deep and Handcrafted Features.
K Suresh Manic1, Venkatesan Rajinikanth2, Ali Saud Al-Bimani1
1National University of Science and Technology, Muscat P.O. Box 112, Oman.
Sensors (Basel, Switzerland)
|January 8, 2023
Summary
This study introduces a novel framework for automatically detecting schizophrenia (SCZ) from brain MRI scans. Combining deep and handcrafted features achieved over 95% accuracy, outperforming other methods for reliable disease screening.
Area of Science:
- Medical Imaging
- Artificial Intelligence in Healthcare
- Neuroscience
Background:
- Brain abnormalities necessitate accurate diagnostic methods.
- Bio-image analysis offers higher accuracy than bio-signal (EEG) analysis for brain condition screening.
- Early and reliable identification of schizophrenia (SCZ) is crucial for patient management.
Purpose of the Study:
- To develop a robust framework for automatic schizophrenia detection using brain MRI slices.
- To enhance diagnostic accuracy by integrating deep and handcrafted features.
- To validate the proposed framework's clinical significance.
Main Methods:
- Utilized VGG16 for deep feature (DF) extraction from MRI slices.
- Collected handcrafted features (HF) and employed the mayfly algorithm for optimal feature selection.
- Concatenated DF and HF, followed by binary classification for SCZ identification.
Main Results:
- The combined DF+HF approach achieved a superior accuracy of >95% for schizophrenia screening.
- Deep features (DF) alone yielded >91% accuracy, while handcrafted features (HF) achieved >85%.
- The proposed framework demonstrated high performance in classifying SCZ from MRI data.
Conclusions:
- The developed framework provides a reliable and accurate method for automated schizophrenia screening from brain MRI.
- Integrating deep and handcrafted features significantly improves diagnostic performance.
- This approach holds clinical significance for future patient screening and diagnosis.
