Radiation Image Denoising Based on Convolutional Neural Network

In this paper, a denoising method for statistical noise in detector based on convolutional neural network was proposed. Using a convolutional neural network model with residual blocks, the method training the radiation image samples in the training dataset and the mapping function of image with noise to image without noise was found. The experiment result shows that the method can reduce the statistical noise while maintaining the image details. The method delivers superior performance in both quantitative parameter and visual feeling compared with other traditional methods.