Key points are not available for this paper at this time.
We develop an algorithm that can detect pneumonia from chest X-rays at a exceeding practicing radiologists. Our algorithm, CheXNet, is a 121-layer neural network trained on ChestX-ray14, currently the largest available chest X-ray dataset, containing over 100, 000 frontal-view-ray images with 14 diseases. Four practicing academic radiologists annotate a set, on which we compare the performance of CheXNet to that of. We find that CheXNet exceeds average radiologist performance on F1 metric. We extend CheXNet to detect all 14 diseases in ChestX-ray14 and state of the art results on all 14 diseases.
Rajpurkar et al. (Tue,) studied this question.