Pith. sign in

REVIEW 2 major objections 4 minor 71 references

An explicit 3D latent volume refined over feed-forward iterations recovers fine CT structure from sparse X-ray views faster and more accurately than prior methods.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.5

2026-07-14 20:50 UTC pith:PJUWMZMA

load-bearing objection Solid engineering paper that puts a persistent 3D latent volume under iterative multi-view refinement; the architecture is coherent and the qualitative gains look real, but the clinical-suitability claim outruns the evidence and the available manuscript is incomplete. the 2 major comments →

arxiv 2603.14915 v2 pith:PJUWMZMA submitted 2026-03-16 cs.CV

ILV: Iterative Latent Volumes for Fast and Accurate Sparse-View CT Reconstruction

Seungryong Lee , Woojeong Baek , Joosang Lee , Eunbyung Park This is my paper
classification cs.CV
keywords sparse-view CTCBCT reconstructionlatent volumefeed-forward 3D reconstructioniterative refinementX-ray feature volumecone-beam CT
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Clinical CT usually needs hundreds of projections, which raises radiation dose, hardware cost, and reconstruction latency. Sparse-view cone-beam CT tries to cut that burden, yet classical algorithms, optimization-based neural fields, and recent feed-forward models still leave streak artifacts and erase fine bone and soft-tissue detail. This paper introduces Iterative Latent Volumes (ILV): a feed-forward network whose core state is an explicit 3D latent volume that is repeatedly updated by multi-view X-ray features and a learned anatomical prior. Supporting modules—an X-ray feature volume, group cross-attention, efficient self-attention, and view-wise aggregation—make that iterative refinement practical. On a dataset of roughly 14,000 CT volumes the method outperforms both prior feed-forward models and optimization-based baselines in quality and speed, supporting the claim that accurate sparse-view CBCT can run at clinically usable latency.

Core claim

The paper claims that maintaining a persistent 3D latent volume and iteratively refining it with multi-view X-ray features plus a data-driven anatomical prior recovers fine structural detail that previous feed-forward sparse-view CBCT models miss, while remaining far faster than per-scene optimization methods, and that this combination is accurate and fast enough for clinical sparse-view reconstruction.

What carries the argument

Iterative Latent Volumes (ILV): an explicit 3D latent volume that is repeatedly updated by injecting a lifted multi-view X-ray feature volume (via group cross-attention) and refined by efficient self-attention and view-wise aggregation, embedding classical iterative-reconstruction principles inside a single feed-forward pass.

Load-bearing premise

A single anatomical prior learned once from the assembled public CT collection will generalize to real clinical sparse-view CBCT acquisitions from different scanners, patient populations, noise levels, and truncation without patient-specific fine-tuning.

What would settle it

Take the trained ILV model, freeze it, and evaluate it without any domain adaptation on a multi-center clinical sparse-view CBCT cohort that uses different scanners, dose protocols, and truncation patterns than the paper’s train/test split; a large drop in PSNR/SSIM or the reappearance of streak and soft-tissue artifacts would falsify the clinical-suitability claim.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Sparse-view CBCT can produce clinically usable volumes at low latency, cutting radiation dose and scanner cost.
  • Feed-forward models that keep a persistent volumetric state outperform direct mappings to triplanes or Gaussian primitives under extreme view sparsity.
  • The same refined latent volume supports both full CT reconstruction and novel-view X-ray synthesis.
  • Time-critical settings such as emergency triage or intraoperative guidance become practical with feed-forward sparse-view reconstruction.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same iterative latent-volume pattern may transfer to other sparse inverse problems such as limited-angle MRI or low-count PET.
  • Real clinical deployment will likely still need domain adaptation or continual prior updates when scanner geometry, noise statistics, or patient demographics differ from the training corpus.
  • Increasing latent resolution or the number of refinement steps may further close residual gaps with dense-view ground truth without leaving the feed-forward regime.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

2 major / 4 minor

Summary. The paper proposes Iterative Latent Volumes (ILV), a feed-forward sparse-view CBCT reconstruction model that maintains an explicit 3D latent volume and iteratively refines it by injecting multi-view X-ray features together with a learned anatomical prior. Architectural components (X-ray feature volume, group cross-attention, efficient self-attention, view-wise aggregation) are introduced to realize this refinement efficiently. The model is trained and evaluated on a large assembled collection of ~14,000 CT volumes and is claimed to outperform both recent feed-forward methods (e.g., DIF-Net, DIF-Gaussian, FreeSeed, X-GRM/X-LRM) and optimization-based NeRF/3DGS approaches in reconstruction quality (PSNR/SSIM-type metrics) and runtime, supporting the claim of clinically usable sparse-view CBCT reconstruction.

Significance. If the quantitative superiority and clinical-transfer claims hold, ILV would be a meaningful advance: it embeds classical iterative-reconstruction principles inside a single feed-forward pass, recovers fine anatomical detail that prior feed-forward models lose, and runs orders of magnitude faster than per-scene optimization methods. The scale of the training corpus (~14k volumes) and the explicit 3D latent-state design are genuine strengths relative to earlier feed-forward CT work. The contribution is therefore of clear interest to the medical-imaging and computational-imaging communities, provided the missing technical and experimental details can be supplied and the generalization premise is stress-tested.

major comments (2)
  1. The manuscript text is severely truncated and garbled: method equations, precise definitions of the latent-volume update, hyper-parameters (iteration count, latent resolution/channels, group size), full metric tables with error bars, and statistical tests are absent or unreadable. Without these, the central claim that ILV 'significantly outperforms' existing methods cannot be independently verified. A complete, self-contained method section and complete quantitative tables are required before acceptance.
  2. The clinical-suitability claim rests on the untested assumption that a single anatomical prior learned from the assembled public CT collection generalizes to real clinical sparse-view CBCT (different scanners, noise statistics, truncation, patient populations) without domain adaptation or patient-specific fine-tuning. No external clinical acquisition, scanner-shift, or noise-robustness experiment is reported. This is load-bearing for the abstract's final sentence and must be addressed, at least by a clear limitation statement and preferably by a transfer experiment.
minor comments (4)
  1. Figure numbering and placement are inconsistent across the supplied text (e.g., Fig. 1 appears early, then Figs. 8–19 appear in an appendix-like block); captions should be self-contained and ordered.
  2. Several related-work citations (X-GRM, X-LRM, recent feed-forward 3DGS CT models) are listed only as arXiv preprints; final versions or DOIs should be supplied when available.
  3. Notation for the latent volume, X-ray feature volume, and the group-cross-attention operator is never formally introduced in the readable portions of the text; a short notation table would help.
  4. The ablation figure (Fig. 9) is described only qualitatively; quantitative ablation numbers should be added to the main results table.

Circularity Check

0 steps flagged

No significant circularity: standard supervised feed-forward architecture trained and evaluated on held-out CT volumes, with no equation or claim reducing a prediction to its own fitted inputs by construction.

full rationale

ILV is an empirical deep-learning paper whose core contribution is an architectural design (explicit 3D latent volume iteratively refined by multi-view X-ray feature injection plus a learned anatomical prior, realized via X-ray feature volume, group cross-attention, efficient self-attention and view-wise aggregation). The claimed superiority is established solely by quantitative reconstruction metrics (PSNR etc.) and qualitative comparisons against baselines on a large public-derived CT collection of ~14k volumes under sparse-view protocols. There are no first-principles derivations, uniqueness theorems, or closed-form predictions whose right-hand side is defined in terms of the left-hand side. Parameters are learned end-to-end from training data and evaluated on held-out volumes; the anatomical prior is the ordinary data-driven prior of supervised learning, not a circular self-definition. Self-citations (e.g., to related feed-forward 3DGS work by overlapping authors) appear only as related-work context and are not load-bearing for any uniqueness or correctness claim. The incomplete/garbled manuscript text does not hide any reduction of the form Eq. X = fitted input; the evaluation remains an independent empirical test. Hence circularity score is zero.

Axiom & Free-Parameter Ledger

3 free parameters · 3 axioms · 3 invented entities

The central claim rests on standard transformer and volume-rendering machinery plus a handful of architectural choices and the untested generalization of a data-driven anatomical prior. No new physical constants or particles are invented; the free parameters are ordinary deep-learning hyper-parameters whose exact values are not even listed in the available text.

free parameters (3)
  • number of latent-volume refinement iterations
    Chosen by the authors to trade quality against speed; the final performance numbers depend on this discrete hyper-parameter.
  • latent volume spatial resolution and channel width
    Architectural sizes that determine both capacity and memory; not derived from first principles.
  • group size in group cross-attention
    Hand-chosen grouping factor that controls the efficiency–expressivity trade-off of the multi-view injection step.
axioms (3)
  • domain assumption Multi-view X-ray features can be lifted into a coherent 3D feature volume via known cone-beam geometry
    Invoked at the start of the method; standard in CBCT but required for the subsequent latent updates to be geometrically meaningful.
  • domain assumption A single anatomical prior learned from the assembled public CT collection is sufficiently representative for sparse-view recovery
    Underpins the claim of clinical suitability; never validated outside the paper’s own distribution.
  • standard math Transformer-style attention (self- and cross-) is a valid mechanism for volumetric feature aggregation
    Background result taken from the vision-transformer literature and used without re-derivation.
invented entities (3)
  • Iterative Latent Volume (ILV) no independent evidence
    purpose: Serves as the persistent 3D reconstruction state that is repeatedly conditioned on multi-view features and the anatomical prior.
    The central architectural invention; no independent physical existence outside the network.
  • X-ray feature volume no independent evidence
    purpose: Intermediate 3D representation that aggregates hybrid 2D X-ray features before injection into the latent volume.
    Paper-specific construct needed to realize the iterative update rule.
  • group cross-attention no independent evidence
    purpose: Efficient mechanism for injecting multi-view information into the latent volume while controlling compute.
    Described as a key architectural component; appears tailored to this paper rather than a previously validated primitive.

pith-pipeline@v1.1.0-grok45 · 13987 in / 2918 out tokens · 39936 ms · 2026-07-14T20:50:42.380282+00:00 · methodology

0 comments
read the original abstract

A long-term goal in CT imaging is to achieve fast and accurate 3D reconstruction from sparse-view projections, thereby reducing radiation exposure, lowering system cost, and enabling timely imaging in clinical workflows. Recent feed-forward approaches have shown strong potential toward this overarching goal, yet their results still suffer from artifacts and loss of fine details. In this work, we introduce Iterative Latent Volumes (ILV), a feed-forward framework that integrates data-driven priors with classical iterative reconstruction principles to overcome key limitations of prior feed-forward models in sparse-view CBCT reconstruction. At its core, ILV constructs an explicit 3D latent volume that is repeatedly updated by conditioning on multi-view X-ray features and the learned anatomical prior, enabling the recovery of fine structural details beyond the reach of prior feed-forward models. In addition, we develop and incorporate several key architectural components, including an X-ray feature volume, group cross-attention, efficient self-attention, and view-wise feature aggregation, that efficiently realize its core latent volume refinement concept. Extensive experiments on a large-scale dataset of approximately 14,000 CT volumes demonstrate that ILV significantly outperforms existing feed-forward and optimization-based methods in both reconstruction quality and speed. These results show that ILV enables fast and accurate sparse-view CBCT reconstruction suitable for clinical use. The project page is available at: https://sngryonglee.github.io/ILV/.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

71 extracted references · 9 linked inside Pith

  1. [1]

    IEEE transactions on medical imaging37(6), 1322–1332 (2018)

    Adler, J., Öktem, O.: Learned primal-dual reconstruction. IEEE transactions on medical imaging37(6), 1322–1332 (2018)

    2018

  2. [2]

    Ultrasonic imaging6(1), 81–94 (1984)

    Andersen,A.H.,Kak,A.C.:Simultaneousalgebraicreconstructiontechnique(sart): a superior implementation of the art algorithm. Ultrasonic imaging6(1), 81–94 (1984)

    1984

  3. [3]

    Biomedical Physics & Engineering Express2(5), 055010 (2016)

    Biguri, A., Dosanjh, M., Hancock, S., Soleimani, M.: Tigre: a matlab-gpu toolbox for cbct image reconstruction. Biomedical Physics & Engineering Express2(5), 055010 (2016)

    2016

  4. [4]

    In: European Conference on Computer Vision

    Cai, Y., Liang, Y., Wang, J., Wang, A., Zhang, Y., Yang, X., Zhou, Z., Yuille, A.: Radiative gaussian splatting for efficient x-ray novel view synthesis. In: European Conference on Computer Vision. pp. 283–299. Springer (2024)

    2024

  5. [5]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Cai, Y., Wang, J., Yuille, A., Zhou, Z., Wang, A.: Structure-aware sparse-view x- ray 3d reconstruction. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 11174–11183 (2024)

    2024

  6. [6]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Caron, M., Touvron, H., Misra, I., Jégou, H., Mairal, J., Bojanowski, P., Joulin, A.: Emerging properties in self-supervised vision transformers. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 9650–9660 (2021)

    2021

  7. [7]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Charatan,D.,Li,S.L.,Tagliasacchi,A.,Sitzmann,V.:pixelsplat:3dgaussiansplats from image pairs for scalable generalizable 3d reconstruction. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 19457– 19467 (2024)

    2024

  8. [8]

    Springer (2024)

    Chen, A., Xu, H., Esposito, S., Tang, S., Geiger, A.: Lara: Efficient large- baselineradiancefields.In:EuropeanConferenceonComputerVision.pp.338–355. Springer (2024)

    2024

  9. [9]

    arXiv preprint arXiv:2303.05754 (2023)

    Chung, H., Lee, S., Ye, J.C.: Decomposed diffusion sampler for accelerating large- scale inverse problems. arXiv preprint arXiv:2303.05754 (2023)

    Pith/arXiv arXiv 2023

  10. [10]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Chung, H., Ryu, D., McCann, M.T., Klasky, M.L., Ye, J.C.: Solving 3d inverse problems using pre-trained 2d diffusion models. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 22542–22551 (2023)

    2023

  11. [11]

    Journal of applied physics34(9), 2722–2727 (1963)

    Cormack, A.M.: Representation of a function by its line integrals, with some radi- ological applications. Journal of applied physics34(9), 2722–2727 (1963)

    1963

  12. [12]

    Cormack, A.M.: Representation of a function by its line integrals, with some radi- ological applications. ii. Journal of Applied Physics35(10), 2908–2913 (1964)

    1964

  13. [13]

    arXiv preprint arXiv:2010.11929 (2020)

    Dosovitskiy, A.: An image is worth 16x16 words: Transformers for image recogni- tion at scale. arXiv preprint arXiv:2010.11929 (2020)

    Pith/arXiv arXiv 2010

  14. [14]

    GitHub3(2019)

    Falcon, W.A.: Pytorch lightning. GitHub3(2019)

    2019

  15. [15]

    Journal of the Optical Society of America A1(6), 612–619 (1984)

    Feldkamp, L.A., Davis, L.C., Kress, J.W.: Practical cone-beam algorithm. Journal of the Optical Society of America A1(6), 612–619 (1984)

    1984

  16. [16]

    Radiology: Artificial Intelligence7(1), e240334 (2024)

    Hermans, S., Hu, Z., Ball, R.L., Lin, H.M., Prevedello, L.M., Berger, F.H., Yusuf, I., Rudie, J.D., Vazirabad, M., Flanders, A.E., et al.: Rsna 2023 abdominal trauma ai challenge: Review and outcomes. Radiology: Artificial Intelligence7(1), e240334 (2024)

    2023

  17. [17]

    description of system

    Hounsfield, G.N.: Computerized transverse axial scanning (tomography): Part 1. description of system. The British journal of radiology46(552), 1016–1022 (1973)

    1973

  18. [18]

    The Lancet Oncology22(4), e136–e172 (2021) 16 S

    Hricak, H., Abdel-Wahab, M., Atun, R., Lette, M.M., Paez, D., Brink, J.A., Donoso-Bach, L., Frija, G., Hierath, M., Holmberg, O., et al.: Medical imaging and nuclear medicine: a lancet oncology commission. The Lancet Oncology22(4), e136–e172 (2021) 16 S. Lee et al

    2021

  19. [19]

    Advances in neural information processing systems 35, 36722–36732 (2022)

    Ji, Y., Bai, H., Ge, C., Yang, J., Zhu, Y., Zhang, R., Li, Z., Zhanng, L., Ma, W., Wan, X., et al.: Amos: A large-scale abdominal multi-organ benchmark for versatile medical image segmentation. Advances in neural information processing systems 35, 36722–36732 (2022)

    2022

  20. [20]

    arXiv preprint arXiv:2505.00702 (2025)

    Jiang, H., Tan, H., Wang, P., Jin, H., Zhao, Y., Bi, S., Zhang, K., Luan, F., Sunkavalli, K., Huang, Q., et al.: Rayzer: A self-supervised large view synthesis model. arXiv preprint arXiv:2505.00702 (2025)

    Pith/arXiv arXiv 2025

  21. [21]

    arXiv preprint arXiv:2505.23716 (2025)

    Jiang, L., Mao, Y., Xu, L., Lu, T., Ren, K., Jin, Y., Xu, X., Yu, M., Pang, J., Zhao, F., et al.: Anysplat: Feed-forward 3d gaussian splatting from unconstrained views. arXiv preprint arXiv:2505.23716 (2025)

    arXiv 2025

  22. [22]

    arXiv preprint arXiv:2410.17242 (2024)

    Jin, H., Jiang, H., Tan, H., Zhang, K., Bi, S., Zhang, T., Luan, F., Snavely, N., Xu, Z.: Lvsm: A large view synthesis model with minimal 3d inductive bias. arXiv preprint arXiv:2410.17242 (2024)

    Pith/arXiv arXiv 2024

  23. [23]

    EBioMedicine62(2020)

    Jin, L., Yang, J., Kuang, K., Ni, B., Gao, Y., Sun, Y., Gao, P., Ma, W., Tan, M., Kang, H., et al.: Deep-learning-assisted detection and segmentation of rib fractures from ct scans: Development and validation of fracnet. EBioMedicine62(2020)

    2020

  24. [24]

    SIAM (2001)

    Kak, A.C., Slaney, M.: Principles of computerized tomographic imaging. SIAM (2001)

    2001

  25. [25]

    arXiv preprint arXiv:2507.23277 (2025)

    Kang, G., Nam, S., Yang, S., Sun, X., Khamis, S., Mohamed, A., Park, E.: ilrm: An iterative large 3d reconstruction model. arXiv preprint arXiv:2507.23277 (2025)

    Pith/arXiv arXiv 2025

  26. [26]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Kang, G., Yoo, J., Park, J., Nam, S., Im, H., Shin, S., Kim, S., Park, E.: Selfsplat: Pose-free and 3d prior-free generalizable 3d gaussian splatting. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 22012–22022 (2025)

    2025

  27. [27]

    ACM Transactions on Graphics42(4) (July 2023),https://repo-sam.inria.fr/fungraph/3d-gaussian-splatting/

    Kerbl, B., Kopanas, G., Leimkühler, T., Drettakis, G.: 3d gaussian splatting for real-time radiance field rendering. ACM Transactions on Graphics42(4) (July 2023),https://repo-sam.inria.fr/fungraph/3d-gaussian-splatting/

    2023

  28. [28]

    In: ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

    Kyung, D., Jo, K., Choo, J., Lee, J., Choi, E.: Perspective projection-based 3d ct reconstruction from biplanar x-rays. In: ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). pp. 1–5. IEEE (2023)

    2023

  29. [29]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Lee, S., Chung, H., Park, M., Park, J., Ryu, W.S., Ye, J.C.: Improving 3d imag- ing with pre-trained perpendicular 2d diffusion models. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 10710–10720 (2023)

    2023

  30. [30]

    In: European Conference on Computer Vision

    Leroy, V., Cabon, Y., Revaud, J.: Grounding image matching in 3d with mast3r. In: European Conference on Computer Vision. pp. 71–91. Springer (2024)

    2024

  31. [31]

    Medical Image Analysis97, 103285 (2024)

    Li, W., Qu, C., Chen, X., Bassi, P.R., Shi, Y., Lai, Y., Yu, Q., Xue, H., Chen, Y., Lin, X., et al.: Abdomenatlas: A large-scale, detailed-annotated, & multi-center dataset for efficient transfer learning and open algorithmic benchmarking. Medical Image Analysis97, 103285 (2024)

    2024

  32. [32]

    In: International Conference on Medical Image Computing and Computer-Assisted Intervention

    Lin, Y., Luo, Z., Zhao, W., Li, X.: Learning deep intensity field for extremely sparse-view cbct reconstruction. In: International Conference on Medical Image Computing and Computer-Assisted Intervention. pp. 13–23. Springer (2023)

    2023

  33. [33]

    In: International Conference on Medical Image Computing and Computer-Assisted Intervention

    Lin, Y., Wang, H., Chen, J., Li, X.: Learning 3d gaussians for extremely sparse- view cone-beam ct reconstruction. In: International Conference on Medical Image Computing and Computer-Assisted Intervention. pp. 425–435. Springer (2024)

    2024

  34. [34]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Lin, Y., Yang, J., Wang, H., Ding, X., Zhao, W., Li, X.: Cˆ 2rv: Cross-regional and cross-view learning for sparse-view cbct reconstruction. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 11205– 11214 (2024) ILV: Iterative Latent Volumes for Sparse-View CT 17

    2024

  35. [35]

    In: International Conference on Medical Image Computing and Computer-Assisted Intervention

    Liu, Y., Yibulayimu, S., Sang, Y., Zhu, G., Wang, Y., Zhao, C., Wu, X.: Pelvic fracture segmentation using a multi-scale distance-weighted neural network. In: International Conference on Medical Image Computing and Computer-Assisted Intervention. pp. 312–321. Springer (2023)

    2023

  36. [36]

    arXiv preprint arXiv:2505.15235 (2025)

    Liu, Y., Li, W., Yu, W., Li, C., Alahi, A., Meng, M., Yuan, Y.: X-grm: Large gaussian reconstruction model for sparse-view x-rays to computed tomography. arXiv preprint arXiv:2505.15235 (2025)

    Pith/arXiv arXiv 2025

  37. [37]

    Journal of the Optical Society of America A37(3), 422–434 (2020)

    Liu, Z., Bicer, T., Kettimuthu, R., Gursoy, D., De Carlo, F., Foster, I.: Tomo- gan: low-dose synchrotron x-ray tomography with generative adversarial networks: discussion. Journal of the Optical Society of America A37(3), 422–434 (2020)

    2020

  38. [38]

    arXiv preprint arXiv:1711.05101 (2017)

    Loshchilov, I., Hutter, F.: Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101 (2017)

    Pith/arXiv arXiv 2017

  39. [39]

    In: International confer- ence on medical image computing and computer-assisted intervention

    Ma, C., Li, Z., Zhang, J., Zhang, Y., Shan, H.: Freeseed: Frequency-band-aware and self-guided network for sparse-view ct reconstruction. In: International confer- ence on medical image computing and computer-assisted intervention. pp. 250–259. Springer (2023)

    2023

  40. [40]

    Scientific data5(1), 1–9 (2018)

    Masoudi, M., Pourreza, H.R., Saadatmand-Tarzjan, M., Eftekhari, N., Zargar, F.S., Rad, M.P.: A new dataset of computed-tomography angiography images for computer-aided detection of pulmonary embolism. Scientific data5(1), 1–9 (2018)

    2018

  41. [41]

    Commu- nications of the ACM65(1), 99–106 (2021)

    Mildenhall, B., Srinivasan, P.P., Tancik, M., Barron, J.T., Ramamoorthi, R., Ng, R.: Nerf: Representing scenes as neural radiance fields for view synthesis. Commu- nications of the ACM65(1), 99–106 (2021)

    2021

  42. [42]

    Organizers, M.C.: Mediastinal lesion analysis (mela) dataset (2022),https:// mela.grand-challenge.org/

    2022

  43. [43]

    Medicina58(2), 241 (2022)

    Otomo, N., Funao, H., Yamanouchi, K., Isogai, N., Ishii, K.: Computed tomography-based navigation system in current spine surgery: a narrative review. Medicina58(2), 241 (2022)

    2022

  44. [44]

    International Journal of Biomedical Imaging2006(1), 010398 (2006)

    Pan, J., Zhou, T., Han, Y., Jiang, M.: Variable weighted ordered subset image reconstruction algorithm. International Journal of Biomedical Imaging2006(1), 010398 (2006)

    2006

  45. [45]

    Peebles,W.,Xie,S.:Scalablediffusionmodelswithtransformers.In:Proceedingsof the IEEE/CVF international conference on computer vision. pp. 4195–4205 (2023)

    2023

  46. [46]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Perera, S., Navard, P., Yilmaz, A.: Segformer3d: an efficient transformer for 3d medical image segmentation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 4981–4988 (2024)

    2024

  47. [47]

    ACM Transactions on Graphics (TOG)41(4), 1–13 (2022)

    Rückert, D., Wang, Y., Li, R., Idoughi, R., Heidrich, W.: Neat: Neural adaptive tomography. ACM Transactions on Graphics (TOG)41(4), 1–13 (2022)

    2022

  48. [48]

    IEEE Transactions on Signal Processing41(2), 534–548 (2002)

    Sauer, K., Bouman, C.: A local update strategy for iterative reconstruction from projections. IEEE Transactions on Signal Processing41(2), 534–548 (2002)

    2002

  49. [49]

    Medical image analysis 42, 1–13 (2017)

    Setio, A.A.A., Traverso, A., De Bel, T., Berens, M.S., Van Den Bogaard, C., Cerello, P., Chen, H., Dou, Q., Fantacci, M.E., Geurts, B., et al.: Validation, com- parison, and combination of algorithms for automatic detection of pulmonary nod- ules in computed tomography images: the luna16 challenge. Medical image analysis 42, 1–13 (2017)

    2017

  50. [50]

    IEEE Transactions on Neural Networks and Learning Systems35(1), 770–782 (2022)

    Shen, L., Pauly, J., Xing, L.: Nerp: implicit neural representation learning with prior embedding for sparsely sampled image reconstruction. IEEE Transactions on Neural Networks and Learning Systems35(1), 770–782 (2022)

    2022

  51. [51]

    Nature biomedical engineering3(11), 880–888 (2019) 18 S

    Shen, L., Zhao, W., Xing, L.: Patient-specific reconstruction of volumetric com- puted tomography images from a single projection view via deep learning. Nature biomedical engineering3(11), 880–888 (2019) 18 S. Lee et al

    2019

  52. [52]

    Physics in Medicine & Biology 53(17), 4777 (2008)

    Sidky, E.Y., Pan, X.: Image reconstruction in circular cone-beam computed tomog- raphy by constrained, total-variation minimization. Physics in Medicine & Biology 53(17), 4777 (2008)

    2008

  53. [53]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Szymanowicz, S., Rupprecht, C., Vedaldi, A.: Splatter image: Ultra-fast single- view 3d reconstruction. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 10208–10217 (2024)

    2024

  54. [54]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Wang, J., Chen, M., Karaev, N., Vedaldi, A., Rupprecht, C., Novotny, D.: Vggt: Visual geometry grounded transformer. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 5294–5306 (2025)

    2025

  55. [55]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Wang, Q., Zhang, Y., Holynski, A., Efros, A.A., Kanazawa, A.: Continuous 3d perception model with persistent state. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 10510–10522 (2025)

    2025

  56. [56]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Wang, S., Leroy, V., Cabon, Y., Chidlovskii, B., Revaud, J.: Dust3r: Geometric 3d vision made easy. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 20697–20709 (2024)

    2024

  57. [57]

    Advances in neural information processing systems34, 12077–12090 (2021)

    Xie, E., Wang, W., Yu, Z., Anandkumar, A., Alvarez, J.M., Luo, P.: Segformer: Simple and efficient design for semantic segmentation with transformers. Advances in neural information processing systems34, 12077–12090 (2021)

    2021

  58. [58]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Xu, H., Peng, S., Wang, F., Blum, H., Barath, D., Geiger, A., Pollefeys, M.: Depth- splat: Connecting gaussian splatting and depth. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 16453–16463 (2025)

    2025

  59. [59]

    In: European Conference on Computer Vision

    Xu, Y., Shi, Z., Yifan, W., Chen, H., Yang, C., Peng, S., Shen, Y., Wetzstein, G.: Grm: Large gaussian reconstruction model for efficient 3d reconstruction and gen- eration. In: European Conference on Computer Vision. pp. 1–20. Springer (2024)

    2024

  60. [60]

    IEEE Transactions on Medical Imaging (2025)

    Yang, J., Shi, R., Jin, L., Huang, X., Kuang, K., Wei, D., Gu, S., Liu, J., Liu, P., Chai, Z., et al.: Deep rib fracture instance segmentation and classification from ct on the ribfrac challenge. IEEE Transactions on Medical Imaging (2025)

    2025

  61. [61]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Yang, J., Sax, A., Liang, K.J., Henaff, M., Tang, H., Cao, A., Chai, J., Meier, F., Feiszli, M.: Fast3r: Towards 3d reconstruction of 1000+ images in one forward pass. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 21924–21935 (2025)

    2025

  62. [62]

    arXiv preprint arXiv:2410.24207 (2024)

    Ye, B., Liu, S., Xu, H., Li, X., Pollefeys, M., Yang, M.H., Peng, S.: No pose, no problem: Surprisingly simple 3d gaussian splats from sparse unposed images. arXiv preprint arXiv:2410.24207 (2024)

    Pith/arXiv arXiv 2024

  63. [63]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Ying, X., Guo, H., Ma, K., Wu, J., Weng, Z., Zheng, Y.: X2ct-gan: reconstructing ct from biplanar x-rays with generative adversarial networks. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 10619– 10628 (2019)

    2019

  64. [64]

    IEEE transactions on medical imaging25(7), 869–881 (2006)

    Yu, L., Zou, Y., Sidky, E.Y., Pelizzari, C.A., Munro, P., Pan, X.: Region of interest reconstruction from truncated data in circular cone-beam ct. IEEE transactions on medical imaging25(7), 869–881 (2006)

    2006

  65. [65]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Zang, G., Idoughi, R., Li, R., Wonka, P., Heidrich, W.: Intratomo: self-supervised learning-based tomography via sinogram synthesis and prediction. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 1960–1970 (2021)

    1960

  66. [66]

    Zha, R., Lin, T.J., Cai, Y., Cao, J., Zhang, Y., Li, H.: R�: Rectifying radiative gaussiansplattingfortomographicreconstruction.arXivpreprintarXiv:2405.20693 (2024)

    Pith/arXiv arXiv 2024

  67. [67]

    In: International Conference on Medical Image Computing and Computer-Assisted Intervention

    Zha, R., Zhang, Y., NAF, H.L.: neural attenuation fields for sparse-view cbct reconstruction. In: International Conference on Medical Image Computing and Computer-Assisted Intervention. pp. 442–452 ILV: Iterative Latent Volumes for Sparse-View CT 19

  68. [68]

    CoRR (2024)

    Zhang, B., Cheng, Y., Yang, J., Wang, C., Zhao, F., Tang, Y., Chen, D., Guo, B.: Gaussiancube: Structuring gaussian splatting using optimal transport for 3d generative modeling. CoRR (2024)

    2024

  69. [69]

    arXiv preprint arXiv:2503.06382 (2025)

    Zhang, G., Zha, R., He, H., Liang, Y., Yuille, A., Li, H., Cai, Y.: X-lrm: X-ray large reconstruction model for extremely sparse-view computed tomography recovery in one second. arXiv preprint arXiv:2503.06382 (2025)

    arXiv 2025

  70. [70]

    In: European Conference on Computer Vision

    Zhang, K., Bi, S., Tan, H., Xiangli, Y., Zhao, N., Sunkavalli, K., Xu, Z.: Gs-lrm: Large reconstruction model for 3d gaussian splatting. In: European Conference on Computer Vision. pp. 1–19. Springer (2024)

    2024

  71. [71]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Ziwen, C., Tan, H., Zhang, K., Bi, S., Luan, F., Hong, Y., Fuxin, L., Xu, Z.: Long- lrm: Long-sequence large reconstruction model for wide-coverage gaussian splats. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 4349–4359 (2025) ILV: Iterative Latent Volumes for Sparse-View CT 3 � ���������� ����������� ������� In this se...

    2025