Deep Learning-Based Inpainting for Chest X-ray Image
Minh-Trieu Tran, Soo-Hyung Kim, Hyung-Jeong Yang, Guee-Sang Lee · 2020
In recent years, chest X-ray images have been progressively applied in research studies. Inspired by the recent success of applying deep learning-based approaches to medical image processing, we first propose an architecture for inpainting on chest X-ray images. A system based on deep convolutional neural networks for completion of the missing or distorted areas using the chest X-ray image was designed and implemented in this paper. Our network was trained with chest X-ray images and shows promising results compared to other networks. Through qualitative and quantitative comparisons with other image inpainting methods, the experimental results have proven our method achieved very good performance when compared with other methods. The average PSNR and SSIM values on the test set for the proposed model were 39.51 dB and 0.79 respectively.