Related Experiment Video
Updated: Oct 1, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.0K
Pediatric age estimation from radiographs of the knee using deep learning
Aydin Demircioğlu1, Anton S Quinsten2, Michael Forsting2
1Department of Diagnostic and Interventional Radiology and Neuroradiology, University Hospital Essen, University of Duisburg-Essen, Hufelandstr. 55, D-45147, Essen, Germany. aydin.demircioglu@uk-essen.de.
European Radiology
|March 2, 2022
Summary
Accurate pediatric age estimation is now possible using knee X-rays and deep neural networks. This method provides reliable chronological age predictions for children and adolescents.
Area of Science:
- Medical imaging analysis
- Artificial intelligence in healthcare
- Pediatric radiology
Background:
- Pediatric age estimation is crucial for forensic, medicolegal, and clinical applications.
- Traditional methods can be invasive or time-consuming.
- Automated age estimation offers a potential solution.
Purpose of the Study:
- To develop and evaluate a deep neural network for automatic chronological age estimation from pediatric knee radiographs.
- To assess the accuracy and reliability of the developed model.
Main Methods:
- A retrospective study utilized 3816 knee radiographs from German pediatric patients (2008-2018) to train a deep neural network.
- The model was validated on an independent cohort of 423 radiographs (2019-2020) and an external cohort of 197 radiographs.
Main Results:
- The deep neural network achieved a mean absolute error of 0.86 ± 0.72 years on the internal validation cohort and 0.9 ± 0.71 years on the external validation cohort.
- The model demonstrated high accuracy in classifying age groups, with AUCs ranging from 0.94 to 0.98 for separating age classes (<14 vs. ≥14 and <18 vs. ≥18 years).
Conclusions:
- Deep neural networks can accurately estimate chronological age in pediatric patients using knee radiographs.
- This AI-driven approach offers a non-invasive and efficient tool for age assessment in clinical and forensic settings.
