How do lower-limb landmarking errors impact morphometrics in total knee arthroplasty planning?

Total knee arthroplasty (TKA) planning relies on accurate anatomical landmark identification to compute morpho-functional metrics guiding implant positioning. While landmark positioning errors inevitably propagate into surgical measurements, current validation practices apply uniform tolerance thresholds regardless of clinical impact. Previous sensitivity analyses identified categorical distinctions and metric-dependent patterns, yet no framework exists for quantifying individual landmark-axis contributions or translating these findings into risk-stratified acceptance criteria.

To establish an evidence-based hierarchy of anatomical landmark criticality for TKA planning through comprehensive sensitivity analysis, and to propose risk-stratified acceptance criteria for quality control protocols.

Monte Carlo-based sensitivity analysis (50,000 iterations per metric) was performed across 30 osteoarthritic patients (67.1 ± 8.8 years) for 14 morpho-functional metrics (11 angles, 3 distances) derived from 34 anatomical landmarks. Manual annotations by two experienced orthopedic surgeons provided empirical positioning error distributions. Atlas-based inverse morphing enabled cross-patient statistical aggregation. Two complementary sensitivity indices were employed: the novel Morphometrical Impact Rate (MIR), quantifying metric deviation per millimeter of landmark displacement, and the Chatterjee coefficient (ξ), capturing non-linear statistical dependencies including reference frame contamination effects.

Metric sensitivity exhibited 12-fold variability (CI₉₀: 1.37°-16.87° for angular metrics). The inferior medial femoral condyle Z-coordinate emerged as the most critical landmark coordinate (MIR = 0.854 deg/mm for angles, 0.792 mm/mm for distances). Malleolar landmarks exhibited maximal sensitivity for rotational metrics (medial malleolus X-axis: MIR = 1.395 deg/mm). Z-axis perturbations dominated coronal measurements, with complete reversal for axial rotational metrics. Plane-defining landmarks demonstrated elevated Chatterjee coefficients despite moderate MIR values, confirming reference frame contamination as a distinct error propagation mechanism.

This study provides quantitative evidence that positioning accuracy and clinical impact are not synonymous. We propose a hierarchical validation framework in which tolerance thresholds scale inversely with quantified clinical impact, prioritizing plane-defining landmarks and high-impact landmark-axis combinations identified through MIR and Chatterjee coefficient analysis. Plane-defining landmarks warrant prioritized validation regardless of individual MIR values, as reference frame errors cascade into multiple geometrically independent metrics.