Related Experiment Videos
Pediatric skeletal age: determination with neural networks
G W Gross1, J M Boone, D M Bishop
1Department of Radiology, Jefferson Medical College, Thomas Jefferson University, Philadelphia, Pa, USA.
Radiology
|June 1, 1995
Summary
A new neural network accurately calculates skeletal age from hand X-rays, performing comparably to expert radiologists. This AI tool may aid in skeletal age assessment.
Area of Science:
- Artificial Intelligence in Medical Imaging
- Pediatric Radiology
- Skeletal Age Assessment
Background:
- Accurate skeletal age assessment is crucial for diagnosing and managing pediatric growth disorders.
- Traditional methods rely on manual interpretation of hand radiographs, which can be subjective and time-consuming.
Purpose of the Study:
- To develop and evaluate a neural network model for automated skeletal age calculation.
- To compare the performance of the neural network against an experienced pediatric radiologist.
Main Methods:
- A neural network was trained using data from 521 digitized hand radiographs.
- Seven linear measurements were used to derive four parameters for network training.
- The jackknife method was employed for model validation.
Main Results:
- The neural network demonstrated a mean difference from biologic age of -0.261 years (SD ±1.82).
- The radiologist showed a mean difference of -0.232 years (SD ±1.54).
- The neural network's skeletal age estimation was closer to biologic age in 47% of cases.
Conclusions:
- A simple neural network can effectively calculate skeletal age from hand radiographs.
- This AI tool shows potential to assist radiologists in skeletal age assessment.
- The developed neural network offers a promising adjunct for clinical practice.