Knee-xRAI: Explainable AI for Accurate Knee Osteoarthritis Grading

Date:

Knee-xRAI: An Explainable AI Framework for Automatic Kellgren-Lawrence Grading of Knee Osteoarthritis

The development of artificial intelligence (AI) in medical imaging has revolutionized various diagnostic processes, yet challenges remain, particularly in the assessment of knee osteoarthritis (KOA). The Kellgren-Lawrence (KL) grading system, widely used for radiographic evaluation of KOA, often suffers from inter-reader variability and lacks transparency in deep learning methodologies. Addressing these issues, researchers have introduced Knee-xRAI, an explainable AI framework designed to enhance the accuracy and interpretability of KL grading.

Overview of Knee-xRAI Framework

Knee-xRAI is a modular framework that independently quantifies three critical radiographic features of KOA:

  • Joint space narrowing (JSN)
  • Osteophytes
  • Subchondral sclerosis

This framework integrates these features into a comprehensive KL grade classification, thereby providing a more detailed and interpretable assessment of knee osteoarthritis.

Technical Components

The Knee-xRAI pipeline employs several advanced techniques:

  • U-Net++ Segmentation: This technique is utilized for contour-based measurement of joint space narrowing.
  • SE-ResNet-50 Network: This component is responsible for per-site osteophyte grading using the OARSI scale.
  • Hybrid Texture-CNN Classifier: This classifier is employed for the binary quantification of subchondral sclerosis.

These components work together to generate a 50-dimensional structured feature vector, which is then processed through two complementary classification paths:

  • XGBoost Path: This path supports SHAP-based feature attribution, enhancing the explainability of the model.
  • ConvNeXt Hybrid Path: By combining the structured feature vector with a full-image encoder, this path aims to improve predictive performance.

Performance Evaluation

The Knee-xRAI framework underwent rigorous evaluation using a dataset derived from the Osteoarthritis Initiative (OAI), comprising 8,260 radiographs. The results highlighted significant achievements:

  • The JSN module attained a Dice coefficient of 0.8909 and an intraclass correlation of 0.8674 for minimum joint space width (mJSW) when compared to manual annotations.
  • The ConvNeXt hybrid path achieved a quadratic weighted kappa (QWK) of 0.8436 and an area under the curve (AUC) of 0.9017, indicating high predictive accuracy.
  • The XGBoost path demonstrated a test QWK of 0.6294, providing full feature-level audit capability.

Insights and Contributions

Ablation studies confirmed that joint space narrowing serves as the dominant predictor in the grading process, achieving a QWK of 0.6103 when considered in isolation. Additionally, osteophyte features provided a consistent yet modest incremental gain of +0.0183, with sclerosis contributing minimally to the overall predictive power. Notably, inference-time ablation of the ConvNeXt hybrid path revealed that the structured pathway significantly enhances performance beyond what the image encoder can achieve alone, as evidenced by QWK drops of 0.098 and 0.284 under different feature manipulation scenarios.

Conclusion

Knee-xRAI represents a significant advancement in the automatic grading of knee osteoarthritis, offering a transparent and auditable approach that quantifies essential radiographic features. This framework not only improves diagnostic accuracy but also fosters trust in AI-driven medical assessments, paving the way for broader applications in healthcare.

Related AI Insights

Lazarus Omolua
Lazarus Omoluahttps://richlyai.com/blog
My mission is to make sure that people in Africa are not left behind in the global AI revolution. RichlyAI exists to give everyone — students, founders, creators, and businesses — the tools to compete globally.

Subscribe

Popular

More like this
Related

How Business Ops Teams Boost Productivity with Codex

Discover how business operations teams use Codex to streamline documentation, enhance collaboration, and improve decision-making with AI-powered automation...

OpenAI Partners with Malta to Offer ChatGPT Plus Nationwide

OpenAI and Malta team up to provide free ChatGPT Plus access and AI training to all citizens, promoting digital literacy and responsible AI use.

Critical Linux Kernel Flaw Risks SSH Host Key Theft

A critical Linux kernel flaw risks stolen SSH host keys. Learn how to protect your systems and stay secure until patches are widely available.

Top External Hard Drives 2026: Expert Reviews & Buying Guide

Discover the best external hard drives of 2026 with expert reviews. Find top picks for speed, durability, and security to suit all storage needs.