In addition, this delineation is prone to inter-expert variability, especially when the images being used for segmentation differ from a center to another (in terms of protocols, modalities and intrinsic MRI quality). This task however requires the delineation of MS lesions: a tedious, manual operation performed by the radiologist. In this context, the number and spread of lesions in the patient’s parenchyma (and their evolution) 2 has become a crucial information on the patient’s disease status, which may then be used for validating the patient treatment.

In this context, Magnetic Resonance Imaging (MRI) plays an important role for the diagnosis 2 and evaluation of the evolution of the disease, thus providing insights to adapt the treatment to each individual due to the highly variable nature of the MS disease course 3.

With the appearance of new treatment molecules modifying the disease evolution (disease modifying drugs - DMD), one of the major challenges in treating multiple sclerosis is now to overcome classical clinical criteria, such as the expanded disability status scale (EDSS), to go towards more sensitive and specific criteria. It is characterized by widespread inflammation, focal demyelination, and a variable degree of axonal loss. Multiple Sclerosis (MS) is a chronic inflammatory disease of the central nervous system affecting around 2.5 million persons worldwide, with a prevalence rate of 00 (higher rates in countries of the northern hemisphere) and a woman:man ratio of around 2.0 1. 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. 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. Each case was annotated manually by an unprecedented number of seven different experts. 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. This allowed for the automatic and independent evaluation of a large range of algorithms in a fair and completely automatic manner.

This challenge was operated using a new open-science computing infrastructure. We present a study of multiple sclerosis segmentation algorithms conducted at the international MICCAI 2016 challenge.

Free instals Pool Challengers 3D