Toward Automatic Articular Cartilage Segmentation From Mri: Distance Constrained Segmentation
by
Feng Huang
Invivo Corporation

Accurate segmentation of articular cartilage is important for MRI-based diagnosis and therapy of cartilage diseases, e.g., osteoarthritis. Fully automatic segmentation is desired since it is both time- and cost-efficient. The automatic segmentation of articular cartilage is challenging because the contrast between cartilage and some surrounding tissues could be too poor to distinguish them. We report a distance constrained method to address the difficulty. In MR images of knee, the contrast between bone and cartilage is usually excellent. Hence the Bone and Cartilage Interface (BCI) is relatively easy to detect. Then the distance from the cartilage surface to the bone is used as a constraint to segment cartilage when contrast cannot provide enough information for segmentation.

Date received: February 20, 2009