Related Experiment Video
Updated: Feb 26, 2026
Software-Assisted Quantitative Measurement of Osteoarthritic Subchondral Bone Thickness
Published on: March 18, 2022
Fully automated subchondral bone segmentation from knee MR images: Data from the Osteoarthritis Initiative
Akash Gandhamal1, Sanjay Talbar2, Suhas Gajre2
1Center of Excellence in Signal and Image Processing, Dept. of Electronics & Telecomm Engineering, SGGS Institute of Engineering & Technology, Nanded, M.S., India; Centre for Intelligent Signal and Imaging Research, Dept. of Electrical and Electronics Engineering, Universiti Teknologi Petronas, Malaysia.
This study presents an automated method for segmenting knee subchondral bone in MR images, crucial for monitoring osteoarthritis (OA) progression. The technique enhances image contrast and uses advanced algorithms, achieving high accuracy for reliable clinical applications.
Area of Science:
- Medical Imaging
- Biomedical Engineering
- Radiology
Background:
- Knee osteoarthritis (OA) progression monitoring relies on subchondral bone changes from MR images.
- Accurate segmentation of subchondral bone is challenging due to image quality and complex anatomy.
- Existing methods struggle with precise bone delineation, hindering reliable OA progression assessment.
Purpose of the Study:
- To develop a fully automated method for segmenting subchondral bone from knee MR images.
- To improve the accuracy and robustness of bone segmentation for OA progression studies.
- To establish a reliable imaging biomarker for monitoring knee OA.
Main Methods:
- Image contrast enhancement using gray-level S-curve transformation.
- Automatic seed point detection via a 3D multi-edge overlapping technique.
- Bone region extraction using distance-regularized level-set evolution and boundary displacement for leakage correction.
Main Results:
- High average sensitivity (91.14% femur, 90.69% tibia) and specificity (99.12% femur, 99.65% tibia).
- Excellent Dice Similarity Coefficient (DSC) scores (90.28% femur, 91.35% tibia).
- Low average surface distance (AvgD) and root mean square surface distance (RMSD) values, indicating minimal error compared to ground truths.
Conclusions:
- The developed automated method demonstrates significant performance, consistency, and robustness in segmenting knee subchondral bone.
- The technique is suitable for large-scale and longitudinal knee OA studies in clinical settings.
- This automated segmentation provides a reliable tool for quantitative assessment of OA progression using imaging biomarkers.
