Review of CNN-Based Approaches for Preprocessing, Segmentation and Classification of Knee Osteoarthritis

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Orthopedics

Background:

• Osteoarthritis (OA) is a common joint disease causing pain and disability.
• Knee osteoarthritis (KOA) is the most frequent type, typically diagnosed via X-ray.
• Current KOA severity classification is subjective, necessitating automated methods.

Purpose of the Study:

• To review and compare Convolutional Neural Network (CNN)-based deep learning approaches for KOA classification.
• To analyze datasets, preprocessing, segmentation, and architectures used in existing studies.
• To identify limitations and suggest future research directions for automated KOA assessment.

Main Methods:

• Systematic literature search across major scientific databases (IEEE Xplore, PubMed, arXiv, etc.) up to March 2025.
• Analysis of studies utilizing deep learning, specifically CNNs, for KOA classification.
• Evaluation of various imaging modalities (X-ray, MRI), preprocessing techniques, and network architectures.

Main Results:

• Deep learning models, particularly CNNs, demonstrate potential for automated KOA classification.
• Reported classification accuracies vary widely (61%–98%) based on data, imaging type, and methodology.
• Significant heterogeneity exists in study designs and reported outcomes.

Conclusions:

• CNN-based deep learning shows promise for objective KOA severity assessment.
• Methodological limitations in current studies hinder clinical translation.
• Future research should focus on enhancing model robustness and generalizability for clinical application.