Haofu Liao

TL;DR: Li et al. as discussed by the authors proposed an unsupervised learning approach to CT metal artifact reduction. But, their method is not suitable for clinical applications and it requires a large amount of synthesized data.

TL;DR: In this article, the authors proposed an end-to-end trainable dual domain network (DuDoNet) to simultaneously restore sinogram consistency and enhance CT images, where the linkage between the sigogram and image domains is a novel Radon inversion layer that allows the gradients to backpropagate from the image domain to the sinogram domain during training.

TL;DR: In this article, a generative encoder-decoder model was proposed to generate radiology reports from chest X-ray images and reports with the following improvements: first, the encoder was pre-trained with a large number of chest images to accurately recognize 14 common radiographic observations, while taking advantage of the multi-view images by enforcing the cross-view consistency.

TL;DR: A generative encoder-decoder model is proposed and extracted medical concepts based on the radiology reports in the training data and fine-tune the encoder to extract the most frequent medical concepts from the x-ray images.

TL;DR: A new topology-adapting deep graph learning approach for accurate anatomical facial and medical landmark detection and quantitative results comparing with the previous state-of-the-art approaches indicating the superior performance in both robustness and accuracy.