Chest CT projections enable better bone suppression in x rays
Anatomy-Decomposed Chest Computed Tomography (CT) Projections as Scalable Supervision for Bone Suppression in Chest Radiographs
Computer Vision and Pattern Recognition
Summary
Bones in chest x rays can hide problems in the lungs, making it harder for doctors to see illnesses. The authors created a way to use 3D chest CT scans to generate fake x rays split into parts: bones, lung tissue, and other soft tissues. Their method lets computer models learn how to remove just the bones from real x rays, even without having pairs of regular and bone-removed images. This helps doctors see lung problems that were previously hidden by bone overlaps.
What this means in practice
- •For radiology software developers: Build tools that improve lung abnormality detection by suppressing bones in chest x rays using models trained on CT-projected synthetic images.
- •For medical ai integration teams: Enhance existing diagnostic AI systems by adding bone-suppression preprocessing without needing paired real training data.
Authors
Mrunmay Angaitkar, Piyush Kumar, Aarjav Satia, Pranav Rao, Ashish Mittal, Manoj Tadepalli, Preetham Putha
Abstract
Bone overlap can obscure abnormalities in chest radiographs, while scarce paired training data limit supervised bone suppression. We address this challenge with a digitally reconstructed radiograph (DRR) framework that converts chest computed tomography (CT) into paired supervision for component suppression. A novel bone segmentation algorithm enables CT decomposition into bone, non-lung soft-tissue, and lung components, which are projected separately. Their weighted combination yields synthetic radiographs with pixel-registered component images that sum exactly to the full DRR. Models trained on these data suppress bone or lung components by predicting the target component and recovering the remainder by subtraction, transferring to real radiographs without real paired training data. As an extension, their outputs on real radiographs provide target domains for unpaired, component-wise DRR translation, reducing the appearance gap while retaining anatomical details. Across multiple public datasets, downstream detection experiments demonstrate the utility of bone suppression, with gains concentrated on abnormalities with substantial bone overlap. Compared with open-source DRR engines applied to the same CTs, our unmodified DRRs achieve comparable realism and preservation of label-relevant anatomy, while translated DRRs achieve the best Fréchet inception distance (FID), lung-field sharpness, and agreement with source-CT anatomy among the evaluated methods. Models and inference code: https://huggingface.co/qureaiorg/bone-suppression; Translated projections: https://huggingface.co/datasets/qureaiorg/ct2xr-projections.