GL-LCM: Global-Local Latent Consistency Models for Fast High-Resolution Bone Suppression in Chest X-Ray Images

¹Hangzhou Dianzi University, ²The Chinese University of Hong Kong, ³University of Electronic Science and Technology of China, ⁴City University of Hong Kong, ⁵Shenzhen University, ⁶Shenzhen Research Institute of Big Data

MICCAI 2025
^†Corresponding Authors

Abstract

Chest X-Ray (CXR) imaging for pulmonary diagnosis raises significant challenges, primarily because bone structures can obscure critical details necessary for accurate diagnosis. Recent advances in deep learning, particularly with diffusion models, offer significant promise for effectively minimizing the visibility of bone structures in CXR images, thereby improving clarity and diagnostic accuracy. Nevertheless, existing diffusion-based methods for bone suppression in CXR imaging struggle to balance the complete suppression of bones with preserving local texture details. Additionally, their high computational demand and extended processing time hinder their practical use in clinical settings. To address these limitations, we introduce a Global-Local Latent Consistency Model (GL-LCM) architecture. This model combines lung segmentation, dual-path sampling, and global-local fusion, enabling fast high-resolution bone suppression in CXR images. To tackle potential boundary artifacts and detail blurring in local-path sampling, we further propose Local-Enhanced Guidance, which addresses these issues without additional training. Comprehensive experiments on a self-collected dataset SZCH-X-Rays, and the public dataset JSRT, reveal that our GL-LCM delivers superior bone suppression and remarkable computational efficiency, significantly outperforming several competitive methods. Our code is available at https://github.com/diaoquesang/GL-LCM.

GL-LCM: Global-Local Latent Consistency Models for Fast High-Resolution Bone Suppression in Chest X-Ray Images

What is bone suppression?

🔻 Try hovering the mouse over the image below and moving it horizontally!

The left side is a Chest X-Ray (CXR) image while the right side is a soft tissue image obtained by bone suppression using our GL-LCM model.

Bone suppression in CXR imaging refers to computational or imaging techniques that minimize or remove the visibility of bone structures from CXR images. The goal is to enhance the clarity of underlying soft tissues, particularly the lungs, to improve diagnostic accuracy for pulmonary conditions.

Abstract

Fig. 1. Overview of GL-LCM framework. (a) Lung segmentation in the pixel space, (b) Dual-path sampling in the latent space, and (c) Global-local fusion in the pixel space.

Fig. 2. Bone suppression and pseudo-color zoomed-in views of GL-LCM compared with state-of-the-art approaches on SZCH-X-Rays (top) and JSRT (bottom) datasets.

Fig. 3. The qualitative results of our ablation study of conditional guidance methods of GL-LCM on the SZCH-X-Rays and JSRT datasets. A pseudo-color zoomed-in view is shown in the bottom right corner, and the green arrows mark the boundary artifacts.

Table 1. Comparison of our method with other universal and task-specifc approaches in the bone suppression task on the SZCH-X-Rays dataset. The best and second-best results are shown in red and blue, respectively.

Table 2. Comparison of our method with other universal and task-specifc approaches in the bone suppression task on the JSRT dataset. The best and second-best results are shown in red and blue, respectively.

Table 3. Inference efficiency comparison of our method and other diffusion-based methods on the SZCH-X-Rays dataset. The best and second-best results are shown in red and blue, respectively.

Table 4. Effect of various conditional guidance methods for local-path sampling on the SZCH-X-Rays and JSRT datasets. The best and second-best results are shown in red and blue, respectively.

Table 5. Effect of various fusion strategies on the SZCH-X-Rays and JSRT datasets. The best and second-best results are shown in red and blue, respectively.

BibTeX