Bone age assessment using convolutional neural networks

In this work, an automated skeletal maturity recognition system is proposed. It first accurately detects the distal radius and ulna (DRU) areas from hand and wrist X-ray images by a faster region-based convolutional neural network model. Then, a well-tuned convolutional neural network (CNN) classification model is applied to estimate the bone ages. We discussed the model performance according to various network configurations. After parameter optimization, the proposed model finally achieved 92% and 88% accuracy for radius and ulna, respectively.