Comprehensive Summary
This study investigates the role of osteophyte formation in knee osteoarthritis (OA) using machine learning (ML)-based quantification of computed tomography (CT) imaging. The goal was to assess osteophyte volume in different anatomical subregions, analyze how medial and lateral OA affects osteophyte distribution, and explore the relationship between osteophytes and OA severity. The dataset included 4,928 CT scans from patients undergoing total or partial knee arthroplasty. The ML-driven imaging analysis segmented osteophytes into femoral and tibial regions, evaluating their 3D volume and distribution. Mean joint space narrowing (JSN) was assessed in medial and lateral compartments, while a Bayesian model was used to analyze osteophyte distribution. Osteophyte volume was correlated with B-scores, a validated measure of OA severity. Key findings revealed that total osteophyte volume strongly correlated between tibia (25%) and femur (75%) (R2 = 0.85). Medial osteophytes (65.3%) were larger than lateral osteophytes (34.6%), and osteophyte distribution remained proportional regardless of total volume size. No significant differences were found between knees with medial, lateral, or both medial and lateral JSN. Additionally, osteophyte volume increased progressively with OA severity, supporting the hypothesis that osteophyte formation represents a “whole-knee” response rather than a localized process. The study underscores the importance of 3D imaging over traditional 2D radiographs, which often fail to detect smaller or obscured osteophytes. Radiographic osteophytes are present in 27-44% of elderly knees, but CT-based analysis provides a more accurate representation of OA progression. The findings suggest that osteophyte analysis could enhance OA diagnostics and surgical planning.
Outcomes and Implications
The study provides critical insights into the role of osteophytes in OA progression and their potential impact on surgical decision-making for knee arthroplasty. The strong correlation between osteophyte volume and OA severity (R2 = 0.85) suggests that osteophytes are an essential marker of disease progression rather than a secondary response to mechanical stress. This has key implications for treatment, as current OA grading systems (such as Kellgren-Lawrence) rely on limited 2D radiographic assessments, which may underestimate disease severity. By using 3D imaging-based osteophyte quantification, clinicians can achieve more precise assessments of knee joint deterioration. The study found that osteophyte volume increases curvilinearly with B-scores, meaning patients with higher osteophyte volumes are more likely to have advanced OA and require total knee arthroplasty (TKA). Additionally, the medial compartment was more affected than the lateral compartment, indicating that medial osteophyte volume may be a key predictor of disease burden. These findings could redefine how OA severity is classified, potentially leading to earlier intervention strategies such as targeted rehabilitation, pharmacological treatments, or knee- preserving surgical techniques. For example, patients with high osteophyte volume but minimal cartilage loss may benefit from unicompartmental knee arthroplasty (UKA) instead of TKA. Furthermore, the study highlights the importance of incorporating AI-driven imaging analysis in routine orthopedic practice. Traditional radiographic analysis of OA lacks precision, while CT- based machine learning models can detect osteophytes with high accuracy. This allows for personalized treatment plans by identifying patients at risk of rapid OA progression. Future research should explore the role of osteophyte volume as a predictor for post-surgical outcomes, particularly in patients undergoing UKA versus TKA. Additionally, further validation of AI models using multicenter datasets will be crucial to refining automated OA severity classification tools. Overall, this study provides a novel framework for using AI-enhanced 3D imaging to quantify osteophyte volume, paving the way for more precise diagnostics, improved surgical planning, and better patient outcomes in knee OA management.