Automatic segmentation of MRI-visible multiple sclerosis (MS) lesions could potentially reduce assessment time and inter- and intra-rater variability. Recently, automatic methods using deep convolutional neural networks (CNN) have obtained great results in image segmentation. This work implements a state-of-the-art 2D CNN-based segmentation method from literature and extends and recalibrates it to a local MS dataset of 91 patients. A clinical evaluation is performed on an independent MS dataset of 53 patients, where 94% of predicted segmentation masks were deemed valuable for clinical use.
This abstract and the presentation materials are available to members only; a login is required.