Spinal Cord Segmentation for Volume Estimation in Healthy and Multiple Sclerosis Subjects Using Crawlers and Minimal Paths
Chris McIntosh, Ghassan Hamarneh, Matthew Toom, Roger C. Tam · 2011
Spinal cord analysis is an important problem in the study of various neurological diseases. Current segmentation and analysis methods in clinical use are slow and labor-intensive, especially for pathological data. "Spinal Crawlers" are a recently developed technique based on an artificial life framework for medical image analysis that complements classical deformable models (snakes and deformable meshes) with high-level control mechanisms. Our method extends Spinal Crawlers to better function in a clinical setting in which images of variable quality and challenging anatomy are encountered. We augment the Spinal Crawler's local optimality with that of globally optimal paths using the live-wire technique. This fusion of globally optimal paths, with locally optimal filtering allows our method to better adapt to contrast changes compared to other methods and therefore allows a larger section of the cord to be measured. Our improvements are validated on 5 vertebral levels of both healthy and pathological spinal cords from clinical MR data. This is the first study to validate a spinal cord segmentation method over a large region encompassing the length of five vertebrae.