List of Papers Browse by Subject Areas Author List
Abstract
Computed tomography (CT) is important in clinical diagnosis, but acquiring high-resolution (HR) CT is constrained by radiation exposure risks. While deep learning-based super-resolution (SR) methods have shown promise for reconstructing HR CT from low-resolution (LR) inputs, supervised approaches require paired datasets that are often unavailable. Zero-shot methods address this limitation by operating on single LR inputs; however, they frequently fail to recover fine structural details due to limited LR information within individual volumes. To overcome these limitations, we propose a novel zero-shot 3D CT SR framework that integrates diffusion-based upsampled 2D projection priors into the 3D reconstruction process. Specifically, our framework consists of two stages: (1) LR CT projection SR, training a diffusion model on abundant X-ray data to upsample LR projections, thereby enhancing the scarce information inherent in the LR inputs. (2) 3D CT volume reconstruction, using 3D Gaussian splatting with our novel Negative Alpha Blending (NAB-GS), which models positive and negative Gaussian densities to learn signed residuals between diffusion-generated HR and upsampled LR projections. Our framework demonstrates superior quantitative and qualitative performance on two public datasets, and expert evaluations present the framework’s clinical potential at 4×.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026/paper/6079_paper.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to the Code Repository
https://github.com/hvcl/3DCT_SR_DDNM_NAB-GS.git
Link to the Dataset(s)
N/A
BibTex
@InProceedings{NohJeo_ZeroShot_MICCAI2026,
author = { Noh, Jeonghyun AND Oh, Hyun-Jic AND Jeong, Won-Ki},
title = { { Zero-Shot CT Super-Resolution Using Diffusion-Based 2D Projection Priors and Signed 3D Gaussians } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 16890},
month = {September},
page = {pending}
}
Reviews
Review #1
- Please describe the contribution of the paper
The paper propose a zero-shot 3D CT SR framework whcih can reformulate volumetric SR as a 3D reconstruction task driven by diffusion-based upsampled 2D projection priors.
- Please list the major strengths of the paper: you should highlight a novel formulation, an original way to use data, demonstration of clinical feasibility, a novel application, a particularly strong evaluation, or anything else that is a strong aspect of this work. Please provide details, for instance, if a method is novel, explain what aspect is novel and why this is interesting.
In this papaer, the null-space based diffusion formulation in [29], is introduced to combine the learned prior with a data-consistency constraint defined by y. At each reverse diffusion step, the estimate is decomposed into range and null-space components. In addition, Negative alpha blending is another point, which can selectively amplify or suppress local intensities of the upsampled LR volume, successfully recovering fine structure details.
- Please list the major weaknesses of the paper. Please provide details: for instance, if you state that a formulation, way of using data, demonstration of clinical feasibility, or application is not novel, then you must provide specific references to prior work.
The proposed method is only compared with methods from 2021 and 2023, making the experimental comparison somewhat insufficient.
- Please rate the clarity and organization of this paper
Satisfactory
- Please comment on the reproducibility of the paper. Please be aware that providing code and data is a plus, but not a requirement for acceptance.
The submission does not provide sufficient information for reproducibility.
- Based on your review and your understanding of the MICCAI Scientific Code of Ethics, do you believe this submission may involve a potential ethics concern or violation?
N/A
- Optional: If you have any additional comments to share with the authors, please provide them here. Please also refer to our Reviewer’s guide on what makes a good review and pay specific attention to the different assessment criteria for the different paper categories: https://conferences.miccai.org/2026/en/REVIEWER-GUIDELINES.html
N/A
- Rate the paper on a scale of 1-6, 6 being the strongest (6-4: accept; 3-1: reject). Please use the entire range of the distribution. Spreading the score helps create a distribution for decision-making.
(4) Weak Accept — marginally above the acceptance threshold, but would not mind if rejected, dependent on rebuttal
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
The designed framework consisting of LR projection SR and 3D CT reconstruction has a certain innovation. Experiments shows superior quantitative and qualitative performance on two public datasets.
- Reviewer confidence
Confident but not absolutely certain (3)
- [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.
N/A
- [Post rebuttal] Please justify your final decision from above.
N/A
Review #2
- Please describe the contribution of the paper
This paper proposes a zero-shot 3D CT super-resolution framework that integrates diffusion-based 2D projection priors with a modified 3D Gaussian splatting reconstruction scheme. The method first employs a diffusion model trained on large-scale X-ray datasets to generate high-resolution projections from low-resolution CT inputs under a DDNM-based formulation. It then reconstructs the final 3D volume using a proposed Negative Alpha Blending Gaussian Splatting (NAB-GS) mechanism that allows signed residual learning by relaxing the non-negativity constraint in standard Gaussian splatting. The approach is evaluated on two public datasets, demonstrating improved quantitative performance compared to existing zero-shot methods and competitive performance relative to supervised approaches.
- Please list the major strengths of the paper: you should highlight a novel formulation, an original way to use data, demonstration of clinical feasibility, a novel application, a particularly strong evaluation, or anything else that is a strong aspect of this work. Please provide details, for instance, if a method is novel, explain what aspect is novel and why this is interesting.
1.The paper considers zero-shot CT super-resolution, which is important in scenarios where paired high-resolution and low-resolution training data are unavailable or difficult to obtain. 2.The Negative Alpha Blending Gaussian Splatting (NAB-GS) formulation enables modeling of signed residual fields, which is a reasonable extension of standard Gaussian splatting and is well motivated for residual-based reconstruction tasks.
- Please list the major weaknesses of the paper. Please provide details: for instance, if you state that a formulation, way of using data, demonstration of clinical feasibility, or application is not novel, then you must provide specific references to prior work.
1.The first stage of the framework relies on a diffusion-based projection enhancement strategy built on established methods such as Denoising Diffusion Null-space Model (DDNM). While the integration into the overall pipeline is reasonable, the methodological contribution of this component appears incremental rather than fundamentally new. 2.While the paper includes several relevant baselines, it does not evaluate against some more recent diffusion-based or hybrid super-resolution models [1]. Including stronger or more up-to-date baselines would provide a more comprehensive assessment of the proposed method’s relative performance. [1] Exploring the feasibility of zero-shot super-resolution in preclinical imaging. MICCAI 2025.
- Please rate the clarity and organization of this paper
Satisfactory
- Please comment on the reproducibility of the paper. Please be aware that providing code and data is a plus, but not a requirement for acceptance.
The submission does not mention open access to source code or data, but provides a clear and detailed description of the algorithm to ensure reproducibility.
- Based on your review and your understanding of the MICCAI Scientific Code of Ethics, do you believe this submission may involve a potential ethics concern or violation?
N/A
- Optional: If you have any additional comments to share with the authors, please provide them here. Please also refer to our Reviewer’s guide on what makes a good review and pay specific attention to the different assessment criteria for the different paper categories: https://conferences.miccai.org/2026/en/REVIEWER-GUIDELINES.html
N/A
- Rate the paper on a scale of 1-6, 6 being the strongest (6-4: accept; 3-1: reject). Please use the entire range of the distribution. Spreading the score helps create a distribution for decision-making.
(4) Weak Accept — marginally above the acceptance threshold, but would not mind if rejected, dependent on rebuttal
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
Please refer to the Strengths and Weaknesses.
- Reviewer confidence
Somewhat confident (2)
- [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.
N/A
- [Post rebuttal] Please justify your final decision from above.
N/A
Review #3
- Please describe the contribution of the paper
1.They propose a new zero-shot 3D CT super-resolution (SR) framework that reformulates volumetric SR as a 3D reconstruction task. This process is driven by 2D projection priors that are upsampled using diffusion models. 2.The authors introduce NAB-GS, a modified version of 3D Gaussian Splatting. By relaxing the traditional non-negativity constraints of standard 3DGS, NAB-GS can accurately learn “signed residual fields” (which inherently contain both positive and negative values) between the diffusion-generated HR projections and the upsampled LR projections.
- Please list the major strengths of the paper: you should highlight a novel formulation, an original way to use data, demonstration of clinical feasibility, a novel application, a particularly strong evaluation, or anything else that is a strong aspect of this work. Please provide details, for instance, if a method is novel, explain what aspect is novel and why this is interesting.
1.The proposed framework is highly efficient compared to existing state-of-the-art zero-shot baselines. 2.The paper goes beyond standard quantitative metrics (PSNR/SSIM) by including an evaluation conducted by domain experts.
- Please list the major weaknesses of the paper. Please provide details: for instance, if you state that a formulation, way of using data, demonstration of clinical feasibility, or application is not novel, then you must provide specific references to prior work.
1.Overly complex pipeline architecture: The proposed pipeline is arguably convoluted, requiring the extraction of 100 LR projections, sequential diffusion inference over every single view (which takes 10 minutes), followed by cubic upsampling, and finally a residual 3DGS optimization phase (taking another 5 minutes) 2.Reconstructing a volume by detouring entirely through a 2D generative process before returning to 3D optimization seems unnecessarily indirect. Directly optimizing a 3D Gaussian Splatting field natively in 3D space—perhaps utilizing a 3D-aware prior or volumetric diffusion model—would likely be a more elegant, streamlined, and computationally efficient approach 3.CuNeRF is a true zero-shot method that relies solely on the internal information of the single provided LR volume. In contrast, the proposed framework utilizes a diffusion model trained on massive external datasets (Chest X-ray14 and CheXpert). This gives the proposed method a massive advantage by injecting external knowledge (priors) that CuNeRF does not have access to. 4.The authors present their method as a highly novel application of 3D Gaussian Splatting for medical super-resolution. However, the literature review and comparative baselines miss highly relevant concurrent advancements in the field. Such as 3DGR-CT.
- Please rate the clarity and organization of this paper
Satisfactory
- Please comment on the reproducibility of the paper. Please be aware that providing code and data is a plus, but not a requirement for acceptance.
The submission does not mention open access to source code or data, but provides a clear and detailed description of the algorithm to ensure reproducibility.
- Based on your review and your understanding of the MICCAI Scientific Code of Ethics, do you believe this submission may involve a potential ethics concern or violation?
N/A
- Optional: If you have any additional comments to share with the authors, please provide them here. Please also refer to our Reviewer’s guide on what makes a good review and pay specific attention to the different assessment criteria for the different paper categories: https://conferences.miccai.org/2026/en/REVIEWER-GUIDELINES.html
N/A
- Rate the paper on a scale of 1-6, 6 being the strongest (6-4: accept; 3-1: reject). Please use the entire range of the distribution. Spreading the score helps create a distribution for decision-making.
(3) Weak Reject — marginally below the acceptance threshold, but would not mind if accepted, dependent on rebuttal
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
the weakness section
- Reviewer confidence
Confident but not absolutely certain (3)
- [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.
N/A
- [Post rebuttal] Please justify your final decision from above.
N/A
Author Feedback
N/A
Meta-Review
Meta-review #1
- Your recommendation
Provisional Accept
- Please justify your decision. In case you deviate from the reviewers’ recommendations, explain in detail the reasons why. In case of an invitation for rebuttal, clarify which points are important to address in the rebuttal.
- Include more comprehensive (stronger and up-to-date) baseline comparisons
- Add more complete literature reviews