By Topic

Automatic Human Knee Cartilage Segmentation From 3-D Magnetic Resonance Images

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

Formats Non-Member Member
$31 $13
Learn how you can qualify for the best price for this item!
Become an IEEE Member or Subscribe to
IEEE Xplore for exclusive pricing!
close button

puzzle piece

IEEE membership options for an individual and IEEE Xplore subscriptions for an organization offer the most affordable access to essential journal articles, conference papers, standards, eBooks, and eLearning courses.

Learn more about:

IEEE membership

IEEE Xplore subscriptions

4 Author(s)
Dodin, P. ; ArthroVision Inc., Montreal, QC, Canada ; Pelletier, J.-P. ; Martel-Pelletier, J. ; Abram, F.

This study aimed at developing a new automatic segmentation algorithm for human knee cartilage volume quantification from MRI. Imaging was performed using a 3T scanner and a knee coil, and the exam consisted of a double echo steady state (DESS) sequence, which contrasts cartilage and soft tissues including the synovial fluid. The algorithm was developed on MRI 3-D images in which the bone-cartilage interface for the femur and tibia was segmented by an independent segmentation process, giving a parametric surface of the interface. First, the MR images are resampled in the neighborhood of the bone surface. Second, by using texture-analysis techniques optimized by filtering, the cartilage is discriminated as a bright and homogeneous tissue. This process of excluding soft tissues enables the detection of the external boundary of the cartilage. Third, a technology based on a Bayesian decision criterion enables the automatic separation of the cartilage and synovial fluid. Finally, the cartilage volume and changes in volume for an individual between visits was assessed using the developed technology. Validation included first, for nine knee osteoarthritis patients, a comparison of the cartilage volume and changes over time between the developed automatic system and a validated semi-automatic cartilage volume system, and second, for five knee osteoarthritis patients, a test-retest procedure. Data revealed excellent Pearson correlations and Dice similarity coefficients (DSC) for the global knee (r = 0.96 , p <; 0.0001, and median DSC = 0.84), for the femur ( r = 0.95, p <; 0.0001, and median DSC = 0.85 ), and the tibia (r = 0.83, p <; 0.0001, and median DSC = 0.84). Very good similarity between the automatic and semi-automatic methods in regard to cartilage loss was also found for the global knee (r = 0.76 and p = 0.016) as well as for the femur (r = 0.79 and p = 0.011). The test-retest reveale- - d an excellent measurement error of -0.3 ±1.6% for the global knee and 0.14 ±1.7% for the femur. In conclusion, the newly developed fully automatic method described herein provides accurate and precise quantification of knee cartilage volume and will be a valuable tool for clinical follow-up studies.

Published in:

Biomedical Engineering, IEEE Transactions on  (Volume:57 ,  Issue: 11 )