Chest x-ray images improved by removing bones using advanced diffusion transformers
DeBoneDiT: Depth-Driven Conditional Bridge Diffusion Transformers for Bone Suppression in Chest X-Ray Images
Computer Vision and Pattern Recognition
Summary
Chest X-Rays often have bones that block the view of important parts, making it harder for doctors to see lung problems. The authors developed a new computer method that uses AI called DeBoneDiT to remove these bones from X-Ray images without needing special machines or extra radiation. Their approach uses a special way to understand image structures and a smart type of AI called diffusion transformers to keep the lung images clear and accurate. They also created a large collection of paired images to help train and test these methods. Their results show this new method works well and is efficient.
What this means in practice
- •For radiology software developers: Develop bone suppression features that enhance lung visibility in chest X-Ray viewers without requiring specialized hardware.$Commercial implications: Enables creation of improved diagnostic imaging tools that reduce reliance on dual-energy machines, expanding product capabilities.
- •For clinical hospital it teams: Integrate AI-based bone suppression to reduce diagnostic uncertainty in chest X-Ray imaging workflows for better pulmonary disease detection.
Authors
Yifei Sun, Fenglei Fan, Junhao Jia, Wenming Deng, Hongxia Xu, Changmiao Wang, Ruiquan Ge
Abstract
Chest X-Ray (CXR) is a primary modality for diagnosing pulmonary diseases, yet the overlap of bony structures often obscures critical radiographic findings, increasing the risk of diagnostic uncertainty. While Dual-Energy Subtraction (DES) is the clinical benchmark for bone suppression, its widespread adoption is primarily hindered by specialized hardware requirements and increased radiation exposure. Recently, diffusion-based generative models have shown promise for this task, yet they are still constrained by reconstruction-biased latent compression, prior-induced domain gap, and a scarcity of high-quality paired datasets. To address these challenges, we propose DeBoneDiT, a depth-driven conditional bridge diffusion Transformer architecture for efficient high-resolution bone suppression. We first introduce Depth Auto-Encoder (DAE), a structure-aware vision tokenizer for latent compression. By incorporating a multi-level depth feature loss derived from the pretrained Depth Anything V2 encoder, DAE preserves both visual fidelity and structural consistency. Second, we formulate bone suppression as a source-conditioned Brownian bridge diffusion process, leveraging CXRs as both the diffusion prior and structural guidance to narrow the source-to-target domain gap and maintain anatomical plausibility. Furthermore, we have constructed and released SZCH-X-Rays, the largest publicly available high-quality paired dataset for bone suppression to date, comprising 741 pairs of CXR and DES soft tissue images. Extensive experiments and downstream evaluation demonstrate that DeBoneDiT achieves superior bone suppression performance with reduced computational overhead.