Artificial intelligence-based thigh muscle volume profiling on CT in elderly East Asian patients with hip fracture
5 affiliations
- 1Department of Orthopaedic Surgery, Daejeon Eulji Medical Center, Eulji University School of Medicine, Daejeon, Republic of Korea
- 2Department of Orthopaedic Surgery, Ajou University School of Medicine, Ajou Medical Center, Suwon, Republic of Korea
- 3Department of Orthopaedic Surgery, Nowon Eulji Medical Center, Eulji University, Seoul, Republic of Korea
- 4Program in Biomedical Science and Engineering, Inha University, Incheon, Republic of Korea
- 5Department of Orthopaedic Surgery, Inha University Hospital, Inha University College of Medicine, Incheon, Republic of Korea
Abstract
Background: Skeletal muscle mass and quality are critical determinants of outcomes in hip fracture patients. However, most prior studies have relied on 2D cross-sectional area measurements, which may not accurately reflect the complex 3D muscle distribution. This study aimed to quantify 3D thigh muscle volume and intramuscular adipose tissue (IMAT) in a large elderly East Asian hip fracture cohort using an AI-assisted segmentation pipeline. Methods: A total of 1,287 patients from a multi-center hip fracture cohort were included. Preoperative whole-thigh CT scans were analyzed using a validated AI-based 3D UNETR segmentation model to delineate 26 individual muscles. Height-normalized muscle volumes (cm³/m²) and IMAT percentages (%) were computed for four anatomic regions: anterior, medial, posterior, and gluteal. Results: The gluteal group exhibited the largest height-normalized muscle volume (855.64 ± 216.38 cm³/m²), followed by the anterior group (784.57 ± 243.17 cm³/m²). On an individual level, the gluteus maximus was the largest single muscle. Paradoxically, the gluteal group showed the highest IMAT percentage (1.43 ± 0.84%), higher than the anterior (0.57 ± 0.42%) and medial groups (0.71 ± 0.51%). Specifically, the highest IMAT percentages were observed in the multifidus (1.93 ± 1.31%) and tensor fasciae latae (1.92 ± 1.03%), which contrasts with the relatively preserved primary locomotive muscles such as the vastus medialis (0.38 ± 0.32%). Conclusions: Elderly hip fracture patients exhibited a dissociation between muscle quantity and quality. The relatively high fatty degeneration observed in stabilizing muscles, including the gluteal muscles and multifidus, despite their large volume, may reflect muscle composition patterns potentially related to postural function. These findings should be considered hypothesis-generating and require validation in prospective studies incorporating direct functional assessments.
Related
- Analysis of barriers to self-monitoring of blood glucose among community-dwelling elderly patients with diabetes mellitus: a cross-sectional study integrating artificial intelligence-based risk prediction.
- From underdiagnosis to scalable screening: Applications of artificial intelligence for vertebral compression fracture detection – a systematic review and meta-analysis of diagnostic performance
- PACS-Integrated Artificial Intelligence-Based Software Tool for Rapid and Fully Automatic Bone Fracture Assessment of the Tibial Plateau
- Falls risk perception and its influencing factors in elderly patients with hip fracture: a cross-sectional study.
- Toward End-to-end Artificial Intelligence Systems for Radiographic Fracture Assessment: A Critical Review of Detection, Explainability, and Prognosis
BibTeX
@article{cha2026artificial,
title = {{Artificial intelligence-based thigh muscle volume profiling on CT in elderly East Asian patients with hip fracture}},
author = {Yong-Han Cha and Jung-Taek Kim and Jin-Woo Kim and Hyun-Bin Kim and Jun-Il Yoo},
journal = {PLOS One},
year = {2026},
month = oct,
doi = {10.1371/journal.pone.0359317},
url = {https://doi.org/10.1371/journal.pone.0359317},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Artificial intelligence-based thigh muscle volume profiling on CT in elderly East Asian patients with hip fracture
AU - Yong-Han Cha
AU - Jung-Taek Kim
AU - Jin-Woo Kim
AU - Hyun-Bin Kim
AU - Jun-Il Yoo
PY - 2026
DA - 2026/10/08
JO - PLOS One
DO - 10.1371/journal.pone.0359317
UR - https://doi.org/10.1371/journal.pone.0359317
AB - Background: Skeletal muscle mass and quality are critical determinants of outcomes in hip fracture patients. However, most prior studies have relied on 2D cross-sectional area measurements, which may not accurately reflect the complex 3D muscle distribution. This study aimed to quantify 3D thigh muscle volume and intramuscular adipose tissue (IMAT) in a large elderly East Asian hip fracture cohort using an AI-assisted segmentation pipeline. Methods: A total of 1,287 patients from a multi-center hip fracture cohort were included. Preoperative whole-thigh CT scans were analyzed using a validated AI-based 3D UNETR segmentation model to delineate 26 individual muscles. Height-normalized muscle volumes (cm³/m²) and IMAT percentages (%) were computed for four anatomic regions: anterior, medial, posterior, and gluteal. Results: The gluteal group exhibited the largest height-normalized muscle volume (855.64 ± 216.38 cm³/m²), followed by the anterior group (784.57 ± 243.17 cm³/m²). On an individual level, the gluteus maximus was the largest single muscle. Paradoxically, the gluteal group showed the highest IMAT percentage (1.43 ± 0.84%), higher than the anterior (0.57 ± 0.42%) and medial groups (0.71 ± 0.51%). Specifically, the highest IMAT percentages were observed in the multifidus (1.93 ± 1.31%) and tensor fasciae latae (1.92 ± 1.03%), which contrasts with the relatively preserved primary locomotive muscles such as the vastus medialis (0.38 ± 0.32%). Conclusions: Elderly hip fracture patients exhibited a dissociation between muscle quantity and quality. The relatively high fatty degeneration observed in stabilizing muscles, including the gluteal muscles and multifidus, despite their large volume, may reflect muscle composition patterns potentially related to postural function. These findings should be considered hypothesis-generating and require validation in prospective studies incorporating direct functional assessments.
SN - 1932-6203
ER -