Related Experiment Video
Updated: Oct 5, 2025
08:51
Magnetic Resonance Imaging of Multiple Sclerosis at 7.0 Tesla
Published on: February 19, 2021
9.3K
A Deep Learning Approach to Predicting Disease Progression in Multiple Sclerosis Using Magnetic Resonance Imaging
Loredana Storelli1, Matteo Azzimonti, Mor Gueye
1From the Neuroimaging Research Unit, Division of Neuroscience.
Investigative Radiology
|January 30, 2022
Summary
A new deep learning algorithm accurately predicts multiple sclerosis (MS) worsening using baseline MRI scans. This AI model outperforms expert physicians in identifying patients at risk of clinical and cognitive decline within two years.
Area of Science:
- Neurology
- Artificial Intelligence
- Medical Imaging
Background:
- Magnetic resonance imaging (MRI) is crucial for multiple sclerosis (MS) management, but its prognostic capability for disease progression remains debated.
- Predicting future disability and cognitive decline in MS patients is essential for personalized treatment strategies.
Purpose of the Study:
- To develop and validate a deep learning algorithm for predicting 2-year clinical and cognitive worsening in MS patients.
- To compare the predictive performance of the deep learning model against expert physician assessments.
Main Methods:
- A convolutional neural network (CNN) architecture was employed using baseline T2-weighted and T1-weighted brain MRI scans from 373 MS patients.
- The CNN model predicted clinical worsening (Expanded Disability Status Scale - EDSS) and cognitive deterioration (Symbol Digit Modalities Test - SDMT), individually and combined.
- Performance was evaluated on an independent dataset and benchmarked against two expert neurologists.
Main Results:
- The CNN model achieved high accuracy in predicting clinical worsening (83.3%) and cognitive deterioration (67.7%).
- Combining EDSS and SDMT data yielded the highest predictive accuracy (85.7%) for the AI model.
- The artificial intelligence approach demonstrated superior performance compared to expert physicians, achieving 85.7% accuracy versus 70% for human raters.
Conclusions:
- A robust deep learning model was developed, capable of accurately predicting 2-year clinical and cognitive worsening in MS patients using baseline MRI.
- This AI tool shows potential to assist clinicians in early identification of MS patients at higher risk of disease progression.
- The algorithm offers a valuable adjunct to conventional MRI analysis for proactive MS patient management.
