MSSEG-2 challenge proceedings: Multiple sclerosis new lesions segmentation challenge using a data management and processing infrastructure

Olivier Commowick (1) , Frédéric Cervenansky (2) , François Cotton (3) , Michel Dojat (4)

1 Empenn - Neuroimagerie: méthodes et applications
2 CREATIS - Centre de Recherche en Acquisition et Traitement de l'Image pour la Santé
3 CHLS - Centre Hospitalier Lyon Sud [CHU - HCL]
4 GIN - [GIN] Grenoble Institut des Neurosciences

This proceedings book gathers methodological papers describing the segmenta-tion methods evaluated at the second MICCAI Challenge on Multiple Sclerosisnew lesions segmentation challenge using a data management and processinginfrastructure. This challenge took place as part of an effort of the OFSEP1(French registry on multiple sclerosis aiming at gathering, for research purposes,imaging data, clinical data and biological samples from the French populationof multiple sclerosis subjects) and FLI2(France Life Imaging, devoted to setupa national distributed e-infrastructure to manage and process medical imagingdata). These joint efforts are directed towards automatic segmentation of MRIscans of MS patients to help clinicians in their daily practice. This challengetook place at the MICCAI 2021 conference, on September 23rd 2021.More precisely, the problem addressed in this challenge is as follows. Con-ventional MRI is widely used for disease diagnosis, patient follow-up, monitoringof therapies, and more generally for the understanding of the natural history ofMS. A growing literature is interested in the delineation of new MS lesions onT2/FLAIR by comparing one time point to another. This marker is even morecrucial than the total number and volume of lesions as the accumulation of newlesions allows clinicians to know if a given anti-inflammatory DMD (disease mod-ifying drug) works for the patient. The only indicator of drug efficacy is indeedthe absence of new T2 lesions within the central nervous system. Performingthis new lesions count by hand is however a very complex and time consumingtask. Automating the detection of these new lesions would therefore be a majoradvance for evaluating the patient disease activity.Based on the success of the first MSSEG challenge, we have organized aMICCAI sponsored online challenge, this time on new MS lesions detection3.This challenge has allowed to 1) estimate the progress performed during the2016 - 2021 period, 2) extend the number of patients, and 3) focus on the newlesions crucial clinical marker. We have performed the evaluation task on a largedatabase (100 patients, each with two time points) compiled from the OFSEPcohort with 3D FLAIR images from different centers and scanners. As in ourprevious challenge, we have conducted the evaluation on a dedicated platform(FLI-IAM) to automate the evaluation and remove the potential biases due tochallengers seeing the images on which the evaluation is made.

Domaines: Imagerie médicale

Olivier Commowick, Frédéric Cervenansky, François Cotton, Michel Dojat. MSSEG-2 challenge proceedings: Multiple sclerosis new lesions segmentation challenge using a data management and processing infrastructure. MICCAI 2021 - 24th International Conference on Medical Image Computing and Computer Assisted Intervention, Sep 2021, Strasbourg, France. , pp.126, 2021. ⟨hal-03358968v3⟩