Accuracy of automated segmentation of gluteus and paraspinal muscles in patients with hip osteoarthritis using TotalSegmentator in 3Dslicer.
Authors
Affiliations (4)
Affiliations (4)
- Department of Orthopaedic Surgery, Yokohama City University, Yokohama, Kanagawa, 236-0004, Japan. [email protected].
- Yokohama City University School of Medicine, Yokohama, Kanagawa, 236-0004, Japan.
- Department of Orthopaedic Surgery, Yokohama City University, Yokohama, Kanagawa, 236-0004, Japan.
- Department of Orthopaedic Surgery, Yokohama City University Medical Center, Yokohama, Kanagawa, 232-0024, Japan.
Abstract
Muscle atrophy is prevalent among patients with hip osteoarthritis, yet quantitative assessment of the peri-hip muscles remains challenging. Although computed tomography (CT) enables objective evaluation, manual segmentation is time-consuming and impractical for routine use. TotalSegmentator, an artificial intelligence-based extension of 3D Slicer, allows rapid automated muscle segmentation. The purpose of this study was to evaluate the accuracy, reproducibility, and clinical relevance of automated peri-hip muscle segmentation using TotalSegmentator in patients with hip osteoarthritis. This retrospective cohort study included 53 female patients (106 hips) with bilateral hip osteoarthritis. Automated segmentation using TotalSegmentator was performed for the gluteus maximus, gluteus medius, and paraspinal muscles and compared with manual segmentation. Muscle volume and CT attenuation were analyzed. Agreement and intraobserver reproducibility were assessed using intraclass correlation coefficients (ICCs). Associations with patient demographics, spinopelvic alignment, and hip function, assessed using the modified Harris Hip Score (mHHS), were also examined in exploratory secondary analyses. TotalSegmentator demonstrated moderate to good reliability for muscle volume and CT attenuation compared with manual segmentation (ICCs 0.47-0.83), with strong correlations between methods (r = 0.79-0.96). Muscle volumes measured by TotalSegmentator were significantly smaller, whereas CT attenuation values were significantly higher than those obtained by manual segmentation. Intraobserver reproducibility was excellent, with ICCs of 1.00 for all muscles. Higher CT attenuation demonstrated a weak positive correlation with young age, better sagittal malalignment, and higher postoperative mHHS. Automated peri-hip muscle segmentation using TotalSegmentator provides rapid, accurate, and reproducible quantitative assessment of muscle volume and quality in patients with hip osteoarthritis.