A deep automated skeletal bone age assessment model via region-based convolutional neural network

Baoyu Liang, Yunkai Zhai, Chao Tong, Jie Zhao, Jun Li, Xianying He, Qianqian Ma

Abstract: Highlights•Faster R-CNN is transferred to a bone age regression model.•The proposed model has a performance of 0.48 MAE and 0.51 MAE on two data sets.•Batch normalization, deeper networks, multi kernel learning and RoI extraction helps.