EXPLORING SEX-BASED HIP MORPHOLOGY DIFFERENCES IN YOUNG ADULTS USING AN AUTOMATED 3D METHOD

INTRODUCTION

Although many factors contribute to the development of hip OA, structural abnormalities such as acetabular dysplasia are recognized as contributing to early degenerative changes. To identify these abnormalities, conventional two-dimensional radiographic methods are still widely used to assess hip morphology. However, they inherently oversimplify the joint’s complex three-dimensional anatomy and are subject to limitations such as patient positioning.

OBJECTIVE

This study aimed to develop and validate an automated method for 3D hip morphology analysis, incorporating segmentation and image feature extraction, as well as to assess morphological differences between sexes.

METHODS

We analyzed data from 2454 participants from Generation R Cohort (mean ± standard deviation age and BMI: 18.4±0.6 years and 22.7±3.8 kg/m²) comprising 1199 males and 1255 females. Hip structures (femoral bone, acetabular bone, femoral cartilage, and acetabular cartilage) were automatically segmented from MRI. An nnU-Net ensemble model was trained on 40 manually segmented hips and its performance evaluated using the Dice score. From the segmentations, five categories of morphological features were computed: basic geometric metrics (centers and radii), cartilage volumes, angular measurements (tilt, version, neck shaft angle and coverage angles, alpha angles), coverage metrics, and joint space width measurements. Independent samples t-tests were used to evaluate sex -based differences.

RESULTS

The automatic deep learning segmentation model achieved mean ± standard deviation Dice scores of 0.97±0.004, 0.90±0.01, 0.76±0.02, and 0.77±0.02 for femoral bone, acetabular bone, femoral cartilage, and acetabular cartilage, respectively, on a hold-out test set of 10 hips. All biomarkers showed statistically significant differences (p<0.05), we highlight those with the largest differences between male and female group means. Compared to males, females had smaller femoral and acetabular radii, as well as reduced cartilage volumes in both the femoral and acetabular regions (Table 1). Alpha angles were lower in females, particularly in the coronal plane, but also in the axial plane, while the acetabular version angle for females was greater (Table 1).

CONCLUSION

This study demonstrates the feasibility of automated 3D analysis for comprehensive hip morphology assessment. Overall, the analysis reveals consistent and measurable differences in hip morphology between sexes. These morphological insights may help clarify structural risk factors for early hip OA.