Related Experiment Video
Updated: Jan 6, 2026
08:52
Software-Assisted Quantitative Measurement of Osteoarthritic Subchondral Bone Thickness
Published on: March 18, 2022
3.4K
Using magnetic resonance imaging-based subregional texture analysis models to classify knee osteoarthritis severity
Keita Nagawa1, Yuki Hara2, Shinji Kakemoto2
1Department of Radiology, Saitama Medical University, 38 Morohongou, Moroyama-machi, Iruma-gun, Saitama, Japan. ldeso5rbdlayids9taiy@gmail.com.
Scientific Reports
|October 16, 2025
Summary
Magnetic resonance imaging (MRI)-based texture analysis (TA) shows potential for classifying knee osteoarthritis (OA) severity by compartment. The medial compartment demonstrated better performance than lateral and patellofemoral compartments in this study.
Area of Science:
- Orthopedics and Imaging Science
- Biomedical Engineering
- Radiology
Background:
- Knee osteoarthritis (OA) is a degenerative joint disease requiring accurate severity assessment.
- Compartmental OA severity grading is crucial for targeted treatment and management.
- Magnetic resonance imaging (MRI) offers detailed joint visualization, but subregional analysis for OA grading needs further exploration.
Purpose of the Study:
- To evaluate the effectiveness of MRI-based subregional texture analysis (TA) models for classifying knee OA severity by compartment.
- To compare the performance of TA models across medial, lateral, and patellofemoral compartments.
- To assess the potential of machine learning algorithms in conjunction with TA for OA grading.
Main Methods:
- 122 knee OA patient MR images (Kellgren-Lawrence grades 2-4) were analyzed using sagittal proton density-weighted and axial fat-suppressed proton density-weighted sequences.
- Texture features were extracted and dimensionally reduced for medial, lateral, and patellofemoral (P-FT) compartments.
- Linear discriminant analysis, support vector machines (SVM), and random forest classifiers were employed and validated using nested cross-validation.
Main Results:
- The developed MRI-based compartmental TA models demonstrated modest performance in classifying OA severity for both individual compartments and the total knee.
- The medial compartment models achieved better classification results compared to the lateral and P-FT compartments.
- Machine learning classifiers showed potential but require further refinement for robust OA severity grading.
Conclusions:
- MRI-based subregional texture analysis holds promise for differentiating knee OA severity grades within specific compartments.
- The medial compartment appears more amenable to TA-based OA severity classification.
- Further research is warranted to optimize TA methods and machine learning algorithms for clinical application in compartmental OA assessment.
