Deep convolutional neural network (CNN) have performed remarkably well on medical image classification, including bone age assessment. The model is designed for evaluating the maturity of a child’s skeletal system using x-ray images. The outstanding CNN model can accurately estimate bone age with a mean average error at 4.49 months. This survey presents existing CNN models for bone age assessment, promising developments, and high-level technical descriptions for implementing the CNN model. It aims to give background knowledge for non-computer scientists interested in deep learning application of bone age assessment. Readers will understand how CNN can improve the performance of bone age assessment models as well as its challenges for future research.
deep learning, convolutional neural network, bone age, growth disorder, maturity estimation, transfer learning
Received: March 25, 2020
Revision received: March 25, 2020
Accepted after revision: April 27, 2020
BKK Med J 2020;16(2): 231-236.
DOI: 10.31524/bkkmedj.2020.23.001