ReCo-Diff: Residual-Conditioned Deterministic Sampling for Cold Diffusion in Sparse-View CT

Yong Eun Choi; Hyoung Suk Park; Kiwan Jeon; Hyun-Cheol Park; Sung Ho Kang

cs.CVMar 3, 2026

ReCo-Diff: Residual-Conditioned Deterministic Sampling for Cold Diffusion in Sparse-View CT

Authors: Yong Eun Choi, Hyoung Suk Park, Kiwan Jeon, Hyun-Cheol Park, Sung Ho Kang

Organizations: National Institute for Mathematical Sciences, Daejeon, 34047, Republic of Korea · Department of Computer Engineering, Korea National University of Transportation, Chungju, 27469, Republic of Korea

Abstract

Cold and generalized diffusion models have recently shown strong potential for sparse-view CT reconstruction by explicitly modeling deterministic degradation processes. However, existing sampling strategies often rely on ad hoc sampling controls or fixed schedules, which remain sensitive to error accumulation and sampling instability. We propose ReCo-Diff, a residual-conditioned diffusion framework that leverages observation residuals through residual-conditioned self-guided sampling. At each sampling step, ReCo-Diff first produces a null (unconditioned) baseline reconstruction and then conditions subsequent predictions on the observation residual between the predicted image and the measured sparse-view input. This residual-driven guidance provides continuous, measurement-aware correction while preserving a deterministic sampling schedule, without requiring heuristic interventions. Experimental results demonstrate that ReCo-Diff consistently outperforms existing cold diffusion sampling baselines, achieving higher reconstruction accuracy, improved stability, and enhanced robustness under severe sparsity.

Figures & tables

Figure 1: Overview of the proposed ReCo-Diff.

Figure 2

Figure 3: Visual comparison of different methods. Reconstructions are obtained with 72 target views and displayed using a window of $[-1000,\,2000]$ Hounsfield Unit (HU). Red regions denote higher errors.

Figure 4: Comparison with CvG-Diff under different SSIM-based reset conditions. CvG-Diff triggers adaptive resetting when the SSIM value falls below a predefined threshold during sampling. In contrast, the proposed method demonstrates a consistently decreasing trend of observation residuals across sampling steps and achieves superior reconstruction quality in practice.

Method (NFE)	18-view			36-view			72-view			Time
	RMSE	PSNR	SSIM	RMSE	PSNR	SSIM	RMSE	PSNR	SSIM
FreeSeed [ 15 ] (1)	50.69	35.49	95.21	23.17	42.28	96.87	18.95	44.03	98.01	0.09
VSS [ 7 ] (1000)	52.73	35.17	90.98	32.87	39.34	95.15	24.19	41.95	97.12	264.71
CoSIGN [ 22 ] (10)	76.95	31.84	86.31	53.73	34.96	89.67	38.37	37.87	93.20	1.82
CvG-Diff [ 3 ] (10)	36.65	38.33	95.18	24.67	41.77	97.05	15.77	45.63	98.54	0.69
ReCo-Diff (16) $\dagger$	37.14	38.21	95.20	23.35	42.22	97.33	14.10	46.60	98.81	0.76

Table 1: Comparison results on the AAPM-LDCT dataset across different view settings. Mean results of RMSE [HU], PSNR [dB], SSIM [%], and average reconstruction time at the 18-view setting [s] are reported. NFE denotes the number of network function evaluations in multi-step reconstruction.

\dagger

indicates an ablation setting where the proposed method is evaluated without the one-time level transition strategy. Lower RMSE and higher PSNR/SSIM indicate better performance. The best result is shown in bold.

Sparse-View CT Reconstruction

Sparse-view computed tomography (CT) reconstruction recovers 3D images from limited X-ray projections while suppressing artifacts and preserving anatomy.

See all 48 papers

Progressively Texture-Aware Diffusion for Contrast-Enhanced Sparse-View CT
Apr 13, 2026Tianqi Wang, Wenchao Du, Hongyu YangSparse-View CT Reconstruction Diffusion-Based Inverse Problems
The College of Computer Science, Sichuan University, Chengdu 610065, China
Conditional Diffusion Posterior Alignment for Sparse-View CT Reconstruction
Apr 23, 2026Luis Barba, Johannes Kirschner, Benjamin BejarCone-Beam CT 3D Medical Imaging
Swiss Data Science Center (SDSC) in Paul Scherrer Institute (PSI), Villigen, Switzerland. · Swiss Data Science Center (SDSC) and ETH Zurich, Switzerland. · Swiss Data Science Center (SDSC) and Paul Scherrer Institute (PSI), Villigen, Switzerland.
LUCID: Learned Undersampling-Adaptive Consistency-Guided Inference with Deterministic Flow Matching for Sparse-View CT Reconstruction
Jun 15, 2026Jigang Duan, Jiayi Wang, Heran Wang +3Sparse-View CT Reconstruction Computational Imaging
School of Mathematical Sciences, Capital Normal University, Beijing 100048, China · National Center for Applied Mathematics Beijing, Capital Normal University and Academy for multidisciplinary studies, Capital Normal University, Beijing, 100048, China

Apr 13, 2026J/K move · Enter open · S save

Progressively Texture-Aware Diffusion for Contrast-Enhanced Sparse-View CT

Tianqi Wang, Wenchao Du, Hongyu Yang

The College of Computer Science, Sichuan University, Chengdu 610065, China

Sparse-View CT Reconstruction Diffusion-Based Inverse Problems Conditional Diffusion Models Diffusion Guidance

Open paper AskPDF arXiv

Diffusion-based sparse-view CT (SVCT) imaging has achieved remarkable advancements in recent years, thanks to its more stable generative capability. However, recovering reliable image content and visually consistent textures is still a crucial challenge. In this paper, we present a Progressively Texture-aware Diffusion (PTD) model, a coarse-to-fine learning framework tailored for SVCT. Specifically, PTD comprises a basic reconstructive module PTD

_{\textit{rec}}

and a conditional diffusion module PTD

_{\textit{diff}}

. PTD

_{\textit{rec}}

first learns a deterministic mapping to recover the majority of the underlying low-frequency signals (i.e., coarse content with smoothed textures), which serves as the initial estimation to enable fidelity. Moreover, PTD

_{\textit{diff}}

aims to reconstruct high-fidelity details for coarse prediction, which explores a dual-domain guided conditional diffusion to generate reliable and consistent textures. Extensive experiments on sparse-view CT reconstruction demonstrate that our PTD achieves superior performance in terms of structure similarity and visual appeal with only a few sampling steps, which mitigates the randomness inherent in general diffusion models and enables a better trade-off between visual quality and fidelity of high-frequency details.

Abstract

Sparse-View CT Reconstruction

Explore similar work

Progressively Texture-Aware Diffusion for Contrast-Enhanced Sparse-View CT

Conditional Diffusion Posterior Alignment for Sparse-View CT Reconstruction

LUCID: Learned Undersampling-Adaptive Consistency-Guided Inference with Deterministic Flow Matching for Sparse-View CT Reconstruction