List of Papers Browse by Subject Areas Author List
Abstract
Clinical pancreatic MRI is central to surveillance for pancreatic ductal adenocarcinoma, yet most clinical scan protocols provide only two approximately orthogonal thick-slice stacks. This yields
anisotropic volumes with limited through-plane resolution, constraining 3D assessment and the visibility of small lesions. Existing slice-to-volume reconstruction (SVR) methods, largely developed for fetal brain MRI, assume multi-stack redundancy and use slice-level outlier rejection, assumptions that do not hold in the low-redundancy two-stack setting where each slice contributes essential information. To address localized cross-view inconsistencies without discarding slices, we propose SRPAN, which introduces pixel-wise reliability weighting via cross-view gating. SRPAN is an implicit neural representation framework that
jointly optimizes per-slice rigid registration and a point spread function (PSF)-aware continuous reconstruction under heteroscedastic uncertainty. In a clinical cohort of 40 high-risk individuals undergoing pancreatic MRI surveillance, SRPAN achieves the highest overall reprojection fidelity (PSNR 25.66, SSIM 0.81, NCC 0.96), leads in 8/9 metrics, and produces the most visually consistent 1 mm isotropic reconstructions across planes. Qualitative results show that SRPAN suppresses artifacts
in regions of cross-view disagreement where symmetric L1 objectives can imprint corrupted structure into the final volume. Code is available at: https://github.com/OlofssonFredrik/SRPAN_MICCAI
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026/paper/1735_paper.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to the Code Repository
https://github.com/OlofssonFredrik/SRPAN_MICCAI
Link to the Dataset(s)
N/A
BibTex
@InProceedings{OloFre_SRPAN_MICCAI2026,
author = { Olofsson, Fredrik AND Wijsman, Koen J. H. AND Engels, Megan M. L. AND Boekestijn, Bas AND Shahbazi Feshtali, Shirin AND van Leerdam, Monique E. AND Broersen, Alexander AND van Hooft, Jeanin E. AND Dijkstra, Jouke},
title = { { SRPAN: Implicit Neural Slice-to-Volume Reconstruction for Super-Resolution Pancreatic MRI } },
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
1.The paper addresses slice-to-volume reconstruction from two approximately orthogonal thick-slice T2-weighted pancreatic MRI stacks in a surveillance setting. 2.The main contribution is an INR-based SVR framework, SRPAN, that jointly models per-slice rigid registration, PSF-aware continuous reconstruction, heteroscedastic uncertainty, and pixel-wise cross-view reliability weighting. 3.The paper claims to improve over prior fetal-MRI-style SVR methods by handling localized cross-view inconsistencies without slice rejection, which is important in a two-stack low-redundancy regime. 4.The evidence presented includes a 40-subject clinical cohort, comparisons against SVRTK, NeSVoR, and a reimplemented SSVR baseline, plus ablations on the key design choices. 5.The practical implication is higher-quality 1 mm isotropic pancreatic MRI reconstructions that may help 3D assessment and future downstream lesion analysis.
- 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 tackles a clinically meaningful and underexplored problem: routine pancreatic MRI often has poor through-plane resolution, and the two-stack setting is genuinely different from the fetal-MRI SVR regime. 2.The method is well matched to the setting; the shift from slice rejection to pixel-wise reliability weighting is sensible for low-redundancy data. 3.The empirical gains are strong on the reported metrics: SRPAN achieves the best overall reprojection fidelity (PSNR 25.66, SSIM 0.81, NCC 0.96) and leads in 8/9 reported metrics across 40 subjects. 4.The ablation study is useful rather than perfunctory; it tests SIREN vs ReLU, PSF modeling, registration, slice scaling, gating, variance clamping, and the loss formulation. 5.The qualitative figures support the claim that the method improves cross-plane consistency and suppresses locally corrupted structure better than symmetric objectives.
- 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 evaluation is indirect: there is no ground-truth isotropic volume, reader study, or downstream task evaluation, so reprojection fidelity alone does not fully establish 3D anatomical correctness or clinical value. 2.Generalization is unclear because the study uses a single internal cohort of 40 high-risk individuals under a specific protocol. 3.The axial-reference gating is protocol-specific and asymmetric; if the axial stack is the noisier one in another setting, the approach may be less reliable. 4.Baseline fairness is somewhat uncertain because SSVR had to be reimplemented by the authors due to lack of official code. 5.Statistical significance testing or confidence intervals for the method comparisons are missing. 6.Training one model per subject may limit practicality, but runtime and compute burden are not discussed.
- Please rate the clarity and organization of this paper
Good
- 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
This paper would be strengthened by: 1.Adding a downstream evaluation, such as pancreas/lesion segmentation or a reader study on lesion conspicuity. 2.Testing on an external cohort or at least on protocol variants where the axial stack is not clearly the cleaner reference. 3.Reporting runtime, memory use, and statistical significance/confidence intervals for the quantitative comparisons. 4.Be open source after being accepted by MICCAI.
- 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.
(6) Strong Accept — must be accepted due to excellence
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
My score is based on the importance of the problem, the reasonable novelty of tailoring INR-based SVR to the two-stack pancreatic setting, and the strength of the reported quantitative and ablation evidence. The main positives are the method-task fit and the sizable gains over the selected baselines. The main concerns are the lack of direct 3D or clinical validation and the limited evidence for generalization beyond one internal cohort. On balance, I place the paper at 6 because it is a meaningful technical contribution with promising results.
- Reviewer confidence
Very confident (4)
- [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 SRPAN, an INR-based method for 3D high-resolution reconstruction of pancreatic MRI using only two thick-slice stacks. The method introduces several core strategies to address the low redundancy inherent in 3D pancreatic super-resolution, including a pixel-wise gating strategy and jointly optimized registration, achieving performance superior to existing 3D reconstruction methods.
- 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 pixel-wise gating strategy enables quality control with only two thick-slice stacks, promoting improved reconstruction quality. (2) The paper demonstrates excellent completeness, with methodological and experimental details presented thoroughly and clearly. (3) The proposed method exhibits strong scalability and holds the potential to become a paradigm for 3D reconstruction tasks involving only two thick-slice stacks. (4) In the absence of ground truth data, the evaluation approach adopted in the paper is reasonably robust and well-justified.
- 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 paper currently ignores in-plane deformations within slices, which may occur in abdominal MRI scans. Relying solely on rigid transformations per slice might not fully correct all motion artifact scenarios. This aspect could be considered in future work. (2) The gating strategy uses the axial plane as the reference. If the axial thick-slice stack suffers from poor scanning quality, the method’s performance may be affected. Enhancing the flexibility of the approach in such cases is advisable.
- Please rate the clarity and organization of this paper
Good
- 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.
(5) Accept — should be accepted, independent of rebuttal
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
The methodological innovation and readability of this paper are both commendable. The proposed method shows great promise as a new paradigm for 3D high-resolution reconstruction under limited thick-slice conditions. Overall, I recommend this paper for acceptance.
- Reviewer confidence
Very confident (4)
- [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
This paper proposes a method for reconstructing isotropic pancreatic MRI volumes from two sets of orthogonal thick-slice acquisitions based on implicit neural representations.
- 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.
The proposed method is sophisticated, taking into account many physical aspects (motion, PSF, view discrepancies). Qualitative results look promising and superior to baseline methods. The ablation study underlines the importance of most of the design choices.
- 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 evaluation is limited based on a single qualitative comparison and quantitative results being limited to re-projection error. Further, reproducibility of the results will be diffiicult given that proprietary data was used and no source code is provided.
- 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?
Hovering between weak accept and weak reject, with the tendency to the former because the approach seems methodogically sound and based in physics. It’s also novel in that it applies SVR to pancreatic MRI.
- 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
Author Feedback
Dear Area Chair and Reviewers,
We sincerely thank you for your time, the constructive feedback, and the recommendation for acceptance. We are thrilled that you recognized the clinical value of addressing low-redundancy pancreatic SVR and the methodological strengths of SRPAN. We would like to address the main points raised in the reviews for the public record:
1.Reproducibility and Source Code (Reviewers 1 & 3) We agree that reproducibility is vital. The full source code for SRPAN is being made open-source, and a link to the GitHub repository will be included in the camera-ready manuscript.
2.Evaluation Metrics and Downstream Validation (Reviewers 1 & 3) We agree that reprojection fidelity is an indirect metric. However, acquiring true isotropic ground-truth MRI is highly challenging, as it is not part of standard clinical care for pancreatic surveillance. We are actively working on obtaining such data for future validation. As noted, evaluating the impact of these reconstructions on downstream clinical and AI tasks is our immediate next step.
3.Rigid Motion Assumption (Reviewer 2) Our dataset was acquired using strict respiratory-triggered and breath-hold protocols. Upon careful inspection of the imaging, we did not observe severe non-rigid artifacts. For this first proposed method, we deliberately chose to keep the registration component simpler and more robust. We agree, however, that extending the framework to handle non-rigid deformations is a valuable future outlook for less strictly controlled acquisitions.
4.Asymmetric Gating Strategy (Reviewers 1 & 2) We agree that fixing the axial stack as the reference could be problematic if it were of poorer quality. While we could manually inspect each case to select the best reference, we fixed the axial stack to ensure the framework remains fully automatic. This choice is grounded in our data’s specific artifact profile: while both scans are highly consistent in the coronal view around the pancreas, the coronal scan exhibits clear image degradations in its out-of-plane (axial) view. Therefore, the axial scan consistently serves as the more reliable reference in our clinical protocol.
Thank you again for your time and expertise.
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.
All 3 reviewers see merit in the paper, particularly highlighting the methodological contributions and the clear way this paper is organized. SVR is applied to the pancreas, a very interesting application with the methodology adapted and made bespoke. 2 recommend clear acceptance, 1 is between weak accept and pending rebuttal - clarifies however that they tend towards accept (this review is also the shortest and rated as not confident).
