We present a study of multiple sclerosis segmentation algorithms conducted at the international MICCAI 2016 challenge. This challenge was operated using a new open-science computing infrastructure. This allowed for the automatic and independent evaluation of a large range of algorithms in a fair and completely automatic manner. This computing infrastructure was used to evaluate thirteen methods of MS lesions segmentation, exploring a broad range of state-of-theart algorithms, against a high-quality database of 53 MS cases coming from four centers following a common definition of the acquisition protocol. Each case was annotated manually by an unprecedented number of seven different experts. Results of the challenge highlighted that automatic algorithms, including the recent machine learning methods (random forests, deep learning, …), are still trailing human expertise on both detection and delineation criteria. In addition, we demonstrate that computing a statistically robust consensus of the algorithms performs closer to human expertise on one score (segmentation) although still trailing on detection scores.

Email address: Olivier.Commowick@inria.fr (Olivier Commowick) Preprint submitted to Nature Scientific Reports July 12, 2018 . CC-BY 4.0 International license It is made available under a (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.The copyright holder for this preprint . http://dx.doi.org/10.1101/367557 doi: bioRxiv preprint first posted online Jul. 13, 2018; We present a study of multiple sclerosis segmentation algorithms conducted at the international MICCAI 2016 challenge. This challenge was operated using a new open-science computing infrastructure. This allowed for the automatic and independent evaluation of a large range of algorithms in a fair and completely automatic manner. This computing infrastructure was used to evaluate thirteen methods of MS lesions segmentation, exploring a broad range of state-of-theart algorithms, against a high-quality database of 53 MS cases coming from four centers following a common definition of the acquisition protocol. Each case was annotated manually by an unprecedented number of seven different experts. Results of the challenge highlighted that automatic algorithms, including the recent machine learning methods (random forests, deep learning, . . . ), are still trailing human expertise on both detection and delineation criteria.In addition, we demonstrate that computing a statistically robust consensus of the algorithms performs closer to human expertise on one score (segmentation) although still trailing on detection scores.

Objectives:To report the clinical, biological, imaging features, and the clinical course of a French cohort of patients with glial fibrillar acidic protein (GFAP) autoantibodies.Methods:We retrospectively included all patients tested positive for GFAP antibodies in the cerebrospinal fluid, by immunohistochemistry and confirmed by cell-based assay using cells expressing human GFAPα, since 2017, from two French referral centers.Results:We identified 46 patients with GFAP antibodies. Median age at onset was 43 years, and 65% were men. Infectious prodromal symptoms were found in 82%. Other auto-immune diseases were found in 22% of patients, and coexisting neural autoantibodies in 11%. Tumors were present in 24%, and T cell dysfunction in 23%. The most frequent presentation was subacute meningoencephalitis (85%) with cerebellar dysfunction in 57% of cases. Other clinical presentation included myelitis (30%), visual (35%) and peripheral nervous system involvement (24%). MRI showed perivascular radial enhancement in 32%, periventricular T2 hyperintensity in 41%, brainstem involvement in 31%, leptomeningeal enhancement in 26%, and reversible splenial lesions in 4 cases. 33/40 patients had a monophasic course, associated to a good outcome at last follow-up (Rankin Score≤2: 89%), despite a severe clinical presentation. Adult and pediatric features are similar. Thirty-two patients were treated with immunotherapy. 11/22 patients showed negative conversion of GFAP antibodies.Interpretation:GFAP auto-immunity is mainly associated with acute/subacute meningoencephalomyelitis with prodromal symptoms, for which tumors and T cell dysfunction are frequent triggers. The majority of patients followed a monophasic course with a good outcome.

No abstract