Assessment of a Clinical Decision Support System (CADIMS) for Improving Multiple Sclerosis Diagnostic Accuracy

Background The results obtained in clinical research led to a new paradigm for the treatment of patients affected with multiple sclerosis (MS): "no evidence of disease activity". This means having each patient monitored with high quality imaging. To this purpose, researchers have developed segmentation algorithms for automated reading of MRIs, facilitating longitudinal comparisons, and allowing an accurate assessment of the number and the volume of focal lesions. We have thus a powerful strategic biomarker for the treatment of MS.

Due to the relevance of comparing over time the evolution of lesions (lesion load) and brain atrophy, the French Society of Neuroradiology edited standardized national recommendations. The MS French Observatory (OFSEP) has adopted these recommendations, in order to harmonize exams for all patients all over the country. At an international level as well, MRI have been put at the heart of the therapeutic strategy.

Therefore, we need now to create the conditions able to implement, collect and analyze imaging data according to these recommendations.

This is the meaning of the MUSIC project, which aims to develop and assess a standardized monitoring MRI control tool for therapeutic decision.

CADIMS is this tool. It is based on an integration of several segmentation algorithms developed by INRIA and an image viewer developed by b<>com and included in a regional images-sharing server ETIAM Nexus. This tool is a help showing segmented brain T2 lesions, gadolinium-enhanced T1 lesions and news T2 lesions appeared from the previous MRI exam, in a multicenter clinical context.

Objective To evaluate the accuracy of MS lesions detection on cerebral MRI by comparing the CADIMS tool to the expert consensus.