CheXseg: Combining Expert Annotations with DNN-generated Saliency Maps for X-ray Segmentation
This addresses the cost and scalability issues in medical image segmentation for healthcare applications, representing a significant but incremental advance over existing methods.
The paper tackles the problem of expensive pixel-level expert annotations for medical image segmentation by proposing CheXseg, a semi-supervised method that combines expert annotations with DNN-generated saliency maps, resulting in a 9.7% improvement over fully-supervised methods and a 73.1% improvement over weakly-supervised methods in mIoU.
Medical image segmentation models are typically supervised by expert annotations at the pixel-level, which can be expensive to acquire. In this work, we propose a method that combines the high quality of pixel-level expert annotations with the scale of coarse DNN-generated saliency maps for training multi-label semantic segmentation models. We demonstrate the application of our semi-supervised method, which we call CheXseg, on multi-label chest X-ray interpretation. We find that CheXseg improves upon the performance (mIoU) of fully-supervised methods that use only pixel-level expert annotations by 9.7% and weakly-supervised methods that use only DNN-generated saliency maps by 73.1%. Our best method is able to match radiologist agreement on three out of ten pathologies and reduces the overall performance gap by 57.2% as compared to weakly-supervised methods.