🟢 Strong Evidence
intelligence/" class="gmj-dict-autolink" title="Dictionary: Artificial Intelligence">Artificial intelligence has revealed cortical lesions in multiple sclerosis patients that were previously undetectable on conventional magnetic resonance imaging (MRI) scans, according to research presented in 2026. The discovery addresses a longstanding clinical blind spot: cortical pathology in MS, which is often the most clinically relevant site of disease, has remained largely invisible to standard imaging techniques used in clinical practice.
Key takeaways
- AI algorithms can detect cortical MS lesions on legacy MRI scans that radiologists cannot see with conventional analysis
- Cortical involvement is often the strongest predictor of disability progression and cognitive decline in MS patients
- The technology works on existing archived MRI data, requiring no new scanning protocols or patient burden
- This capability could change how MS is monitored and how treatment decisions are made in clinical practice
The Cortical Lesion Detection Gap in MS Imaging
Comparison of lesion visibility: conventional MRI versus AI-enhanced analysis
Source: MS imaging research, 2026 | Georgian Medical Journal News
Why Cortical Lesions Matter in MS
Cortical gray matter pathology in MS has long been recognized by neuropathologists as a hallmark of the disease, yet it has remained largely invisible to clinicians using standard MRI protocols. This disconnect between pathological reality and clinical visibility has created a significant diagnostic and prognostic challenge. Research in MS journals shows that cortical involvement correlates more strongly with disability progression and cognitive decline than white matter lesion burden alone.
The reason for this invisibility is technical: conventional MRI sequences are optimized to detect white matter lesions, which have high contrast against surrounding tissue. Cortical lesions, by contrast, blend into adjacent gray matter and are largely isointense on standard T2-weighted and proton-density sequences. This means patients can have substantial cortical pathology without it appearing on the MRI images their neurologists review.
How AI Makes Invisible Lesions Visible
The AI systems being deployed for this task use convolutional neural networks trained on large datasets of MRI scans paired with manual expert annotations and, in validation studies, neuropathological confirmation. These algorithms learn spatial and intensity patterns that distinguish cortical MS lesions from normal-appearing cortex, even when those patterns are too subtle for human radiologists to detect reliably. The technology works on legacy MRI data—scans that are already in patients’ medical records—without requiring new imaging sequences or additional radiation exposure.
Critically, the AI approach requires no changes to how patients are scanned or how institutions manage their MRI protocols. This dramatically lowers the barrier to implementation: existing archives of MRI scans can be retrospectively analyzed, and future scans can be processed immediately after acquisition. For a healthcare system managing thousands of MS patients, this means potential access to previously hidden prognostic information at minimal cost and no additional patient burden.
Clinical and Prognostic Implications
If validated in prospective clinical trials, AI-detected cortical lesions could reshape how MS activity is assessed and monitored. Currently, neurologists rely on relapse history, disability scores, and white matter lesion counts to gauge disease activity and guide treatment escalation. Adding cortical lesion burden to this calculus could improve prediction of which patients are at highest risk for disability progression. Studies on MS cognition suggest cortical involvement is a major driver of cognitive dysfunction, a symptom that significantly impacts quality of life but is often underestimated in routine clinical assessments.
The detection of previously invisible cortical lesions also raises questions about treatment adequacy in current MS populations. Patients thought to have stable or slowly progressing disease on conventional MRI might actually have accumulating cortical pathology. This could justify earlier or more aggressive therapy in patients who appear well-controlled by traditional metrics but have significant AI-detected cortical burden.
Cortical lesions detected by AI on standard MRI scans show spatial and temporal patterns consistent with active MS pathology, suggesting they represent genuine disease activity rather than imaging artifacts.
— Multiple sclerosis imaging research teams, 2026
What this means
Road to Clinical Adoption
While the technical capability to detect cortical lesions is now demonstrated, several steps remain before this becomes standard clinical practice. Published guidance on AI validation in medical imaging emphasizes the need for external validation across multiple healthcare institutions, demonstration of clinical utility in prospective patient cohorts, and integration into institutional workflows. Regulatory approval from agencies such as the FDA or EMA may be required depending on how the AI tool is deployed and marketed.
Early adopter institutions are likely to test cortical lesion detection in research settings before widespread clinical rollout. This allows for assessment of inter-reader variability, reproducibility across different MRI manufacturers and field strengths, and most importantly, correlation with clinical outcomes. Understanding which patients benefit most from cortical lesion information—and which treatment decisions are actually changed by knowing cortical burden—is essential before recommending routine use.
Frequently asked questions
Are cortical lesions in MS different from white matter lesions?
Yes. Cortical lesions are pathologically distinct: they often extend into the adjacent white matter and meninges, and they tend to show less inflammation than white matter lesions. However, cortical lesions correlate more strongly with disability and cognitive decline than white matter lesion count alone, suggesting they reflect more severe or progressive pathology.
Will AI-detected cortical lesions change how my MS is treated?
Possibly, but this depends on whether formal clinical trials demonstrate that knowledge of cortical burden improves treatment decisions and patient outcomes. Current evidence supports an association between cortical involvement and worse outcomes, but prospective studies are needed to show that detecting cortical lesions early and intensifying treatment based on that finding actually improves long-term disability. Your neurologist will ultimately decide whether to use cortical lesion information in your care.
Can cortical lesions be reversed with current MS treatments?
There is limited evidence that existing disease-modifying therapies significantly reverse established cortical pathology, though some studies suggest higher-efficacy agents may slow cortical lesion accumulation. This is an active area of research. Emerging research on neuroprotective strategies aims to repair cortical damage, but these are not yet standard clinical treatments.
The ability to detect cortical lesions in MS using artificial intelligence represents a significant technical advance with potential to reshape how the disease is monitored and treated. Whether this capability translates into better patient outcomes will depend on rigorous clinical validation and thoughtful integration into existing care models. The next phase of research must move beyond proving that AI can see what humans cannot, to demonstrating that what AI sees actually matters for predicting disease course and guiding therapeutic decisions.
Source: AI unlocks previously invisible cortical lesions in MS using legacy MRI scans
Was this article helpful?
Disclaimer. This article is health journalism intended for general information and education. It is not medical advice and is not a substitute for professional diagnosis or treatment. Always consult a qualified healthcare provider about your individual circumstances. Full disclaimer →
Related Coverage
Editorial standards. This article was produced under the GMJ News editorial process, with oversight by the GMJ Editorial Board. Our editorial process. Spotted an error? Contact the editorial team.
