Section: New Results
Medical Image Computing in Multiple Sclerosis
Automatic Segmentation of lesions and Normal Appearing Brain Tissues (NABT) in patients with Multiple Sclerosis
I set up a workflow for the automatic segmentation of Multiple Sclerosis lesions in MR images. This work consisted, in first place, in compare different algorithms of registration, inhomogeneity correction, denoising to normalize and enhance MR images and select the best workflow for this concrete application. In second place, to improve an algorithm of segmentation develop in our laboratory, STREM, to handle different sequences and multicenter studies. This work has been performed in collaboration with Dr. JC Ferré from the neuroradiology department of University Hospital of Rennes.
Automatic Characterisation of the Normal Appearing White Matter (NAWM) in Multiple Sclerosis
The purpose of this work was to study how people in the literature cope with the issue of quantitative analysis of Normal Appearing White Matter tissues from MRI by using either standard weighted MRI sequences or relaxometry sequences. The outcome of this study is that quantitative MRI requires true T1/T2 parameters estimation but this may be done in spite of making these new sequences hard to normalize for multi centric experiments and hard also to set up in clinical conditions.