Automatic segmentation of the spinal cord and intramedullary multiple sclerosis lesions with convolutional neural networks
The spinal cord is frequently affected by atrophy and/or lesions in multiple sclerosis (MS) patients. Segmentation of the spinal cord and lesions from MRI data provides measures of atrophy and lesion burden, which are key criteria for the diagnosis, prognosis and longitudinal monitoring in MS. Achieving robust and reliable segmentation across multi-center spinal cord data is challenging because of the large variability related to acquisition parameters and image artifacts. In particular, a precise delineation of lesions is hindered by a broad heterogeneity of lesion contrast, size, location, and shape. The goal of this study was to develop a fully-automatic framework, robust to variability in both image parameters and clinical condition, for segmentation of the spinal cord and intramedullary MS lesions from conventional MRI data. 1042 adult subjects (459 healthy controls, 471 MS patients, and 112 with other spinal pathologies) were included in this multi-center study (n=30 centers). Data spanned 3 contrasts (T1-, T2- and T2*-weighted) for a total of 1943 volumes and featured large heterogeneity in terms of resolution, orientation, coverage and clinical conditions. The proposed cord and lesion automatic segmentation approach is based on a cascade of two Convolutional Neural Networks (CNN): a first CNN with 2D dilated convolutions detects the spinal cord centerline followed by a second 3D CNN that segments the spinal cord and lesions. When compared to a state-of-the-art spinal cord segmentation method (PropSeg), our CNN-based approach showed a median Dice of 95 lesion segmentation, our framework, when compared with manual segmentation of MS patients, provided a lesion-wise detection sensitivity of 83 of 77 framework is open-source and available in the Spinal Cord Toolbox.
READ FULL TEXT