Femorotibial degeneration equilibrium analysis for objective and clinically interpretable knee osteoarthritis stratification.

Abstract

Radiographic evaluation of knee osteoarthritis (KOA) is subject to inter-observer variability and limited objectivity, particularly in early disease stages. This study aimed to develop and validate a clinically interpretable framework for objective KOA identification and severity stratification by quantifying anatomical and functional degenerative imbalance from routine knee radiographs. We propose a novel Femorotibial Degeneration Equilibrium Analysis (FDEA) framework that models KOA as a progressive biomechanical imbalance process rather than an appearance-based classification task. Patient-specific anatomical stability anchors were first extracted to establish a femorotibial reference framework. A virtual mechanical load axis was then projected to quantify medial-lateral load drift, while joint space integrity was assessed using continuity tracing without pixel-wise segmentation. Directional degeneration gradients were further estimated to capture compartment-specific disease progression. These clinically meaningful indicators were integrated using orthopaedic reasoning rules to derive an objective equilibrium loss score for KOA severity stratification. The proposed method was evaluated using knee radiographs from the Osteoarthritis Initiative (OAI) dataset with expert Kellgren-Lawrence (KL) grading as the reference standard and was compared with joint space width measurement, texture-based analysis, shape deformation analysis, and radiologist rule-based scoring. External validation was performed on an independent subset from the Multicenter Osteoarthritis Study (MOST). FDEA demonstrated superior agreement with expert KL grading (Cohen's κ = 0.86 ± 0.04; p < 0.01) and achieved higher sensitivity for early-stage KOA (92.4% ± 3.1%) and specificity for advanced disease (94.1% ± 2.8%) than all comparative methods. Assessment time did not differ significantly from conventional radiographic evaluation. The proposed FDEA framework offers a transparent, clinically interpretable decision-support tool that complements conventional grading and supports early diagnosis and treatment planning in knee osteoarthritis.