Related Experiment Video
Updated: Jun 26, 2025
Magnetic Resonance Imaging of Multiple Sclerosis at 7.0 Tesla
Published on: February 19, 2021
Diagnostic effectiveness of deep learning-based MRI in predicting multiple sclerosis: A meta-analysis
Tareef S Daqqaq1, Ayman S Alhasan1, Hadeel A Ghunaim1
1From the Department of Internal Medicine (Daqqaq, Alhasan, Ghunaim),College of Medicine, Taibah University, Madinah, and from Department of Radiology (Daqqaq), Prince Mohammed Bin Abdulaziz Hospital, Ministry of National Guard Health Affairs, and from the Department of Radiology (Alhasan), King Faisal Specialist Hospital and Research Center, Madinah, Kingdom of Saudi Arabia.
Convolutional neural networks (CNN) applied to MRI scans show high accuracy in identifying, classifying, and segmenting multiple sclerosis (MS) lesions. This automated approach offers a promising tool for prompt and effective MS diagnosis.
Area of Science:
- Neurology and Medical Imaging
- Artificial Intelligence in Medicine
- Machine Learning for Disease Diagnosis
Background:
- Multiple sclerosis (MS) is an inflammatory disease affecting the central nervous system (CNS).
- Early detection of MS lesions is crucial for effective management.
- Convolutional Neural Networks (CNNs) offer pattern recognition capabilities for medical image analysis.
Purpose of the Study:
- To evaluate the diagnostic performance of CNN-based MRI for MS lesion identification, classification, and segmentation.
- To assess the accuracy, sensitivity, and specificity of CNN models in MS diagnosis.
- To determine the effectiveness of 2D-3D CNNs in analyzing brain MRIs for MS.
Main Methods:
- A systematic literature search was conducted across multiple databases (PubMed, Web of Science, Embase, Cochrane Library, CINAHL, Google Scholar).
- Studies reporting the use of CNN-based MRI for MS diagnosis were included.
- Performance metrics evaluated included accuracy, sensitivity, specificity, and Dice Similarity Coefficient (DSC).
Main Results:
- The 2D-3D CNN demonstrated high accuracy (98.81%), sensitivity (98.76%), and specificity (98.67%) in identifying MS lesions.
- Classification accuracy was also significantly high at 91.38%.
- The Dice Similarity Coefficient (DSC) for lesion segmentation was 63.78%, indicating strong performance.
Conclusions:
- 2D-3D CNN-based MRI represents a robust, automated system for MS diagnosis.
- This technology shows high diagnostic performance and potential for prompt and effective disease prediction.
- The findings support the integration of CNNs into clinical practice for MS management.
