Osteoarthritis (OA) is the most common form of joint disease and a major cause of long-term disability in the United States (US). It is estimated that 2.5% of the adult population have symptomatic knee or hip OA. Over two-thirds of the 7.8 million OA patients in the US who seek treatment have moderate to severe joint involvement and would benefit from a therapy which arrests or delays cartilage loss. The etiology of OA is still partially unclear: While genetic factors are believed to underlie a significant proportion of OA cases, the majority of occurrences may not be genetically predetermined. OA is influenced by diet, body condition, or physical stress experienced (due to injury or overuse of a joint). Patient condition may therefore likely be improved or further progression prevented by an early identification of OA progression, combined with effective therapies. However, the current armamentarium of OA therapies merely relieves the inflammation and painful symptoms of OA but does not suppress the ongoing degenerative process. There is no known cure for osteoarthritis and further drug research is essential to help OA patients.

Cartilage loss is believed to be the dominating factor in OA. While the standard radiography-based analysis method relies on joint-space width as a surrogate measure for cartilage thickness, an increasing body of literature supports the use of MRI as a primary imaging method to evaluate progression of osteoarthritis. MRI is able to directly measure cartilage volume and thickness. Being a three-dimensional imaging modality it allows, unlike x-ray projection images, for a localized analysis of imaging data in the full three-dimensional spatial context. Significant advances in MRI have resulted in the ability to quantify cartilage morphology and thereby provide a means to evaluate potential effects of pharmacologic intervention on OA progression.

To aid drug development and to help subsequent regulatory approval, accurate, quantitative methods are needed to rapidly screen MR imaging data. We are therefore working on fully-automatic image analysis methods to allow for the alignment of knee MR images, and for bone and cartilage segmentation and quantification. [1] [2] [3] [4]

We developed the Knee Segmentation and Registration Toolkit (KSRT). This software can be downloaded from here.


  1. Automatic Atlas-based Three-label Cartilage Segmentation from MR Knee Images,
    Shan, L., Charles C., and Niethammer M.
    , Proceedings of the Workshop on Mathematical Methods in Biomedical Image Analysis (MMBIA), (2011)
  2. Automatic three-label bone segmentation from knee MR images,
    Shan, L., Zach C., and Niethammer M.
    , International Symposium on Biomedical Imaging (ISBI), (2010)
  3. Automatic Bone Segmentation and Alignment from MR knee images,
    Shan, L., Zach C., Styner M., Charles C., and Niethammer M.
    , SPIE Medical Imaging, (2010)
  4. Globally Optimal Finsler Active Contours,
    Zach, C., Shan L., Frahm J. - M., and Niethammer M.
    , DAGM, p.552–561, (2009)