TP4.8 Automated 3D Hip Landmark Detection Using Supine and Weight-Bearing CT Imaging: A Validation Study

Clicks: 10
ID: 330954
2026
Free PDF
Article Quality & Performance Metrics
Overall Quality
Not rated
Combines reader engagement with the AI quality analysis. This article has not been analysed, so there is no overall score — reader engagement is measured and shown alongside.
AI Quality Assessment
Not analyzed
Readership in this journal
Popular

Ranked #341 of 444 articles by views in Journal of hip preservation surgery

Most read Least read

Bar heights use a square-root scale. Only the 120 most-read articles are drawn; the journal has 444 in total.

Mint this article as an NFT
Not yet minted

Create a permanent, verifiable on-chain record of this article on the Scimatic Network. The NFT is held in your Journament account, and you can withdraw it to your own wallet at any time.

5 SUSD one-off · no wallet required
Sign in to mint
Abstract
Abstract Background Accurate identification of pelvic and proximal femoral landmarks is essential for CT-based hip assessment, diagnosis, and preoperative planning. However, manual three-dimensional landmarking is time-consuming and subject to observer variability. Automated landmark detection may improve efficiency, standardisation, and reproducibility. Hypothesis An artificial intelligence–based workflow combining automated segmentation with statistical shape model–based landmark transfer can identify clinically relevant hip landmarks on CT with performance comparable to manual interobserver variability. Study design Retrospective validation study. Methods The automated segmentation model used in this workflow was previously developed and trained on 722 CT scans. For the present validation study, 30 CT scans were included: 20 conventional angio-CT scans and 10 weight-bearing CT scans. The pelvis, femur, and sacrum were automatically segmented using this deep learning model. A statistical shape model–based non-rigid fitting procedure was then used to transfer predefined anatomical landmarks to each subject-specific three-dimensional bone model. All models were expressed in a standardised anterior pelvic plane coordinate system. Three observers independently annotated the landmarks on the segmented bone models. Automated landmark locations were compared with manual annotations using three-dimensional Euclidean distance error, and interobserver variability was calculated using pairwise Euclidean distances between observers. Results Automated landmark detection showed close agreement with manual annotations across the evaluated landmarks. In most cases, automated–observer differences were of similar magnitude to the natural interobserver variability. Mean automated errors ranged from 0.00 ± 0.00 mm to 7.61 ± 2.51 mm, while interobserver variability ranged from 0.00 ± 0.00 mm to 7.25 ± 3.40 mm. These findings indicate observer-level performance within the studied cohort and imaging conditions. Conclusion This AI-based workflow demonstrated observer-level performance for automated hip landmark detection. The method appears promising as a standardised and observer-independent tool for CT-based hip morphometric assessment and preoperative planning. Further validation in larger and more heterogeneous populations is needed. Clinical Relevance Automated hip landmark identification may reduce manual workload and improve reproducibility in CT-based planning and three-dimensional hip morphometric analysis. Key Terms deep learning, hip alignment, medical image analysis, statistical shape modelling, computed tomography.
Reference Key
openalex_W7220367242 Use this key to autocite in the manuscript while using SciMatic Manuscript Manager or Thesis Manager
Authors Ide Van den Borre, MD Jasper Lambrechts, MD Aline Van Oevelen, BSc Emile Versluys, Robin Vanhauwe, PhD Emmanuel Audenaert MD
Are you one of these authors? Claim your profile
Journal Journal of hip preservation surgery
Year 2026
DOI
10.1093/jhps/hnag043.161
URL
https://doi.org/10.1093/jhps/hnag043.161
Keywords Keywords not found

Citations

No citations found. To add a citation, contact the admin at info@scimatic.org

Comments

No comments yet. Be the first to comment on this article.