Automating the diagnosis of MS

Multiple sclerosis (MS) is an autoimmune disease characterized by the demyelination of the central nervous system. Traditionally, diagnosing MS relies on clinical symptoms, laboratory tests, and MRI scans. However, recent advancements in imaging technology and artificial intelligence (AI) are changing the landscape of MS diagnosis.

Optical coherence tomography (OCT) is a key technology that reflects early changes in retinal layer thickness, indicating potential retinal nerve fiber atrophy. This is crucial as thinner retinal layers often correlate with worse visual outcomes. AI is now being integrated with OCT to enhance diagnostic accuracy.

In a recent study published in Translational Vision Science & Technology, researchers explored the independent use of infrared reflectance scanning laser ophthalmoscopy (IR-SLO) images to automate MS diagnosis.

“To our knowledge, this is a pioneering study incorporating IR-SLO into automated diagnosis of MS…among the few studies that have applied deep learning to retinal imaging data for detecting MS.”

About the study

The study involved 32 MS patients and 70 healthy controls, using multiple convolutional neural networks (CNNs) trained on both OCT and IR-SLO data. The results demonstrated a bimodal model that significantly outperformed models trained on either image type alone, achieving 92% accuracy and 95% sensitivity.

Interestingly, the model’s performance was validated with an external dataset, yielding an accuracy of 85% and 99.7% for both AUROC and AUPRC curves, indicating a robust diagnostic tool.

Conclusions

While traditional OCT-based models have their limitations, combining OCT and IR-SLO data enhances diagnostic accuracy by 3%. This hybrid approach showcases the potential of AI in identifying subtle changes in retinal imaging that may be overlooked by human evaluators.

As the field progresses, larger and more diverse studies will be necessary to further validate these findings and integrate them into clinical practice.

For more information, refer to the original article: AI and imaging advances revolutionize multiple sclerosis diagnosis.