Researchers have developed a new artificial intelligence model capable of identifying "invisible" cortical lesions in patients with multiple sclerosis (MS) using standard clinical MRI scans. This tool, detailed in a study published in Radiology, allows clinicians to detect these subtle brain abnormalities that frequently elude human observation, potentially offering a more accurate assessment of disease progression and disability.
Detecting Subpial Lesions with Artificial Intelligence
Cortical lesions, particularly those located in the subpial region—the outermost layer of the brain’s gray matter—are a hallmark of multiple sclerosis but are notoriously difficult to visualize on conventional MRI sequences. According to research led by teams at the Johns Hopkins University School of Medicine, these lesions are closely linked to physical and cognitive decline in MS patients.
The AI model functions by analyzing 3D T1-weighted and FLAIR MRI scans. By automating the detection of these previously "invisible" markers, the technology reduces the reliance on highly specialized, time-consuming ultra-high-field MRI scanners, which are not available in most clinical settings. The researchers trained the algorithm on a diverse dataset to ensure it could generalize across different scanner manufacturers and magnetic field strengths.
Clinical Implications for MS Management
The ability to quantify cortical lesion load in a standard clinical workflow changes how neurologists might monitor MS. Traditional imaging often focuses on white matter lesions, yet grey matter damage is a primary driver of progressive disability.
By identifying these lesions earlier and more reliably, healthcare providers can better distinguish between different stages of the disease. This is particularly relevant for patients who experience worsening symptoms despite a stable appearance of white matter lesions on standard imaging. The integration of this AI tool into radiological software may assist in more personalized treatment planning, allowing for earlier intervention with disease-modifying therapies.
Comparison of Imaging Modalities
The shift toward AI-enhanced standard MRI represents a move toward greater accessibility in MS care.
| Feature | Standard MRI (Human Review) | AI-Enhanced MRI Analysis |
|---|---|---|
| Detection of Subpial Lesions | Very Low | High |
| Requirement for Specialized Scanners | Often required for clarity | Not required |
| Consistency | Variable (Reader dependent) | High (Automated) |
| Clinical Workflow Speed | Moderate | Fast |
Limitations and Future Directions
While the results are promising, the research team emphasizes that the model is designed to support, not replace, the clinical judgment of radiologists. The performance of the AI is dependent on the quality of the input images, and further validation across broader, multi-center populations is necessary before widespread adoption.
Future studies are expected to focus on how this automated detection correlates with long-term clinical outcomes. As the technology matures, it could become a standard component of neuroimaging reports, providing a more comprehensive view of the structural damage caused by multiple sclerosis.