Related Experiment Video
Updated: Apr 12, 2026
07:56
Scanning Skeletal Remains for Bone Mineral Density in Forensic Contexts
Published on: January 29, 2018
17.7K
Automated system of analysis to quantify pediatric hip morphology
Cliodhna N Gartland1, John Healy2, Rosanne-Sara Lynham3
1Insight Research Ireland Center for Data Analytics, University College Dublin, Belfield, D04C1P1, Dublin, Ireland; University College Dublin, Belfield, D04C1P1, Dublin, Ireland.
Computers in Biology and Medicine
|August 29, 2025
Summary
This study developed an AI system to precisely identify 22 hip landmarks on X-rays, aiding in the objective assessment of developmental dysplasia of the hip (DDH) for earlier diagnosis.
Area of Science:
- Radiology
- Medical Imaging
- Artificial Intelligence
Background:
- Developmental dysplasia of the hip (DDH) lacks objective diagnostic metrics.
- Accurate diagnosis is crucial for timely intervention.
Purpose of the Study:
- To develop an AI system for precise anatomical landmark detection in juvenile hip radiographs.
- To derive novel morphological measures for DDH assessment.
Main Methods:
- Implemented a coarse-to-fine U-Net deep neural network approach.
- Compared six coarse and four fine model variations, optimizing data augmentation, input size, attention gates, and loss functions.
Main Results:
- Achieved a 3.79 mm root-mean-square error in landmark detection accuracy.
- Positional accuracy demonstrated low bias and precision comparable to clinical experts.
Conclusions:
- The developed AI system accurately detects key hip landmarks.
- This system shows potential for creating objective assessment metrics for DDH.
