DeVAR: Low-Dose CT Denoising via Visual Autoregressive Modeling

2026-06-26

AI Summary

This work addresses the challenge of severe noise and artifacts in low-dose computed tomography (LDCT) images, which compromise diagnostic accuracy due to reduced radiation exposure. The authors propose DeVAR, a novel framework that introduces visual autoregressive modeling to LDCT denoising for the first time, progressively generating normal-dose CT images through prefix-guided discrete tokens. To mitigate information loss caused by discrete quantization, they design a residual refiner to recover high-frequency details and develop a dual-representation hybrid training strategy that effectively integrates continuous and discrete latent variables. Evaluated on two public datasets, DeVAR demonstrates significant improvements over existing methods in both quantitative metrics and visual quality.

Abstract

Computed tomography (CT) plays a crucial role in medical diagnosis, but minimizing radiation exposure while maintaining image quality remains a critical challenge. Low-dose CT (LDCT) protocols reduce radiation risks but inevitably suffer from severe noise and artifacts that compromise diagnostic accuracy. While existing deep learning methods have achieved promising results, there remains a continuous quest for generative paradigms that intrinsically capture global-to-local structural dependencies to better preserve fine anatomical details. To this end, we propose DeVAR, a novel generative framework that applies visual autoregressive modeling (VAR) to LDCT denoising for the first time. Conditioned on global context provided by LDCT prefix tokens, DeVAR progressively generates discrete token maps of the target normal-dose CT (NDCT) via next-scale prediction. Because quantization inherently discards high-frequency information, we introduce a residual refiner to capture subtle anatomical structures beyond the capacity of a discrete codebook. Finally, empowered by a dual-representation hybrid training strategy, our hybrid NDCT decoder seamlessly integrates continuous and discrete latents to reconstruct high-fidelity, detail-preserved images. Extensive experiments on two public datasets demonstrate that DeVAR consistently achieves superior qualitative and quantitative performance compared to state-of-the-art LDCT denoising methods.

Problem

• Low-Dose CT
• Denoising
• Image Quality
• Radiation Exposure
• Artifacts

Innovation

• visual autoregressive modeling
• low-dose CT denoising
• discrete token generation
• residual refiner
• hybrid training strategy