Quantitative assessment of disease progression in multiple sclerosis using MRI is an important issue for therapeutic monitoring and understanding to the development of disability. In particular, the relationship between spinal cord atrophy and development of disability has been the subject of recent research interest. We aim to segment the spinal cort from MR images and have a complete describtion of its surface in order to provide atrophy measurements from this surface, such as cross-sectional area. The proposed method is bases on geodesic active contour, with surface represented implicitly so that we do not need any parameterization. Some performance approaches are proposed to enhance the segmentation methods performance. In this paper, we included the image's local information around the evolved contour to discriminate the spinal cord from the other organ. The method helps the contour active evolution stops exactly on the spinal cord surface desired to be extracted. quantitatively, segmentation result obtained using this model is better than those proposed previously. The needed time to perform this segmentation is also shorter Keywords: Multiple sclerosis, Deformable model, Contour Active Geodesic, IRM, Segmentation Citra