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. To be time- and cost-effective, computer-assisted 3D image analysis is essential. However, most image-analysis methods for OA still require significant human intervention, precluding the comprehensive analysis of large databases as for example acquired by the Osteoarthritis Initiative. A strategy that has been beneficial in studies of the brain is the use of atlases to assist in data analysis. Following such success we propose the creation of population-based bone and cartilage atlases to facilitate bone and cartilage segmentation and to allow for localized data analysis by representing imaging data in a common anatomical coordinate system. We will use the developed methods to analyze cartilage thickness and to perform correlations with clinical variables. Developed software tools will be distributed in open-source form.

Public Health Relevance

Osteoarthritis (OA) is a debilitating disease, with millions of people affected in the US alone. While large-scale OA studies by the Osteoarthritis Initiative and Pfizer have acquired a breathtaking wealth of data, a comprehensive analysis of the imaging data has proven difficult, due to the unavailability of robust, fully-automatic computer analysis methods. This project will develop such automatic image analysis tools to allow for the efficient extraction of quantitative measures from magnetic resonance images, to ultimately aid drug development to help patients afflicted by a disabling disease without a current cure.

National Institute of Health (NIH)
National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS)
Exploratory/Developmental Grants (R21)
Project #
Application #
Study Section
Special Emphasis Panel (ZRG1-SBIB-J (80))
Program Officer
Lester, Gayle E
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
University of North Carolina Chapel Hill
Biostatistics & Other Math Sci
Schools of Arts and Sciences
Chapel Hill
United States
Zip Code
Huang, Chao; Shan, Liang; Charles, H Cecil et al. (2015) Diseased Region Detection of Longitudinal Knee Magnetic Resonance Imaging Data. IEEE Trans Med Imaging 34:1914-27
Shan, Liang; Zach, Christopher; Charles, Cecil et al. (2014) Automatic atlas-based three-label cartilage segmentation from MR knee images. Med Image Anal 18:1233-46
Huang, Chao; Shan, Liang; Charles, Cecil et al. (2013) Diseased region detection of longitudinal knee MRI data. Inf Process Med Imaging 23:632-43
Shan, Liang; Charles, Cecil; Niethammer, Marc (2012) AUTOMATIC MULTI-ATLAS-BASED CARTILAGE SEGMENTATION FROM KNEE MR IMAGES. Proc IEEE Int Symp Biomed Imaging 2012:1028-1031
Shan, Liang; Charles, Cecil; Niethammer, Marc (2012) Automatic Atlas-based Three-label Cartilage Segmentation from MR Knee Images. Proc Workshop Math Methods Biomed Image Analysis :241-246