Abstract
Introduction
The alpha angle remains the standard metric for quantifying cam deformity in femoroacetabular impingement syndrome (FAIS), although this metric reduces a three-dimensional deformity to a two-dimensional angular measurement. A three-dimensional volumetric approach offers superior reliability and clinical correlation during femoroplasty. The purpose of this study is to three-dimensionally phenotype cam deformity to best target surgical resection and characterize the breadth of FAIS morphology.
Methods
After processing 2 GB of data across 156 radial CT scans of patients indicated for hip arthroscopy for FAIS at a tertiary referral center, we trained an nnU-Net segmentation model on 56 manually annotated hip CTs and applied it to a separate cohort of 100 FAIS patients indicated for arthroscopy. Two fellowship-trained hip specialists virtually performed femoroplasty with an agreed upon planned resection. For each patient, we automatically isolated the femoral head, fit a best-fit sphere through iterative refinement, and extracted morphological descriptors spanning four domains: head size (fitted sphere diameter), cam deformity (4 deviation metrics), resection geometry (volume, depth, surface area, solid angle), and spatial localization (resection centroid, extent, and span along the proximal-distal axis of the femoral neck). We performed Pearson correlation analyses, principal component analysis (PCA), and k-means clustering to characterize the morphology-resection relationship. We acquired pre- and postoperative patient-reported outcome scores (mHHS).
Results
K-means clustering yielded three cam phenotypes (Figure 1A, Figure 2): Type A/Hillock (33%), characterized by focal, proximal, gentle elevations necessitating relatively shallow resections closer to the femoral head; Type B/Hill (24%), characterized by broader surface coverage reflective of femoral head size with intermediate elevation; and Type C/Mountain (43%), characterized by distal, extreme elevations with intermediate surface area. Resection breadth depended upon head size (diameter vs. surface area r = 0.67, p < 0.001; Figure 1B), while resection depth was tightly coupled with distal extension (depth vs. centroid distal position r = 0.87, p < 0.001; Figure 1C). Breadth and depth were only moderately correlated (surface area vs. max depth r = 0.50; Figure 1D), confirming these as semi-independent dimensions. Patient-reported outcome scores did not differ significantly across phenotypes at any postoperative timepoint (all p > 0.05).
Discussion
Surgical resection in cam FAIS is governed by two orthogonal dimensions: head size determines cam surface area, while cam elevation is tightly coupled with distal cam extent. Three distinct three-dimensional cam phenotypes exist (Hillock, Hill, and Mountain) and establish a foundation for three-dimensional diagnostic evaluation and arthroscopic management, enabling heightened diagnostics beyond the alpha angle and individualized preoperative planning to best execute the femoroplasty.