Related Experiment Video
Updated: Nov 25, 2025
05:05
Four-Dimensional CT Analysis Using Sequential 3D-3D Registration
Published on: November 23, 2019
8.3K
Automated age estimation of young individuals based on 3D knee MRI using deep learning
Markus Auf der Mauer1, Eilin Jopp-van Well2, Jochen Herrmann3
1Medical and Industrial Image Processing, University of Applied Sciences of Wedel, Feldstraße 143, 22880, Wedel, Germany. markusalexander.adm@gmail.com.
International Journal of Legal Medicine
|December 17, 2020
Summary
This study introduces an automated machine learning method for age estimation using 3D knee MRIs. The approach offers a non-invasive, objective alternative for forensic medicine, achieving high accuracy in age regression and classification.
Area of Science:
- Forensic Medicine
- Medical Imaging
- Machine Learning
Background:
- Accurate age estimation is vital in forensic medicine, especially for individuals lacking documentation.
- Current methods are often labor-intensive, subjective, and may involve radiation.
- Non-invasive magnetic resonance imaging (MRI) shows potential for correlating bone ossification with chronological age.
Purpose of the Study:
- To develop a fully automated, computer-based method for age estimation using 3D knee MRIs.
- To address the need for user-independent approaches in large-scale datasets.
- To utilize machine learning for reliable and objective age assessment.
Main Methods:
- A three-part solution involving image-preprocessing, bone segmentation, and age estimation.
- Utilized 3D knee MRI volumes (coronal and sagittal) from 185 and 404 Caucasian male subjects, aged 13-21.
- Employed a combination of convolutional neural networks and tree-based machine learning algorithms.
Main Results:
- Achieved a mean absolute error of 0.67 ± 0.49 years in age regression.
- Demonstrated classification accuracy of 90.9% with 88.6% sensitivity and 94.2% specificity (18-year age limit).
- Results highlight the potential of deep learning for automated age estimation.
Conclusions:
- The developed automated method provides a reliable and objective approach for age estimation from 3D knee MRIs.
- This non-invasive technique can overcome limitations of traditional forensic age assessment methods.
- Further improvements are anticipated with larger and more diverse training datasets.
