Pith. sign in

REVIEW 3 major objections 5 minor 28 references

RadGS-Reg: Registering Spine CT with Biplanar X-rays via Joint 3D Radiative Gaussians Reconstruction and 3D/3D Registration

T0 review · 3 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read RadGS-Reg claims that spine CT/X-ray registration is best done in 3D: reconstruct a radiative-Gaussians volume from two X-rays, then regress its 6-DOF pose against the CT—1.14 mm error, 93.33% success, 0.82 s.

desk verdict Useful idea, honest ablation, but the third-stage pretraining on the test fold's CT is a real evaluation flaw; the headline numbers need a fair baseline comparison before being called SOTA. read the letter →

arxiv 2508.21154 v1 pith:Z6HIMFNI submitted 2025-08-28 cs.CV cs.AI

Ao Shen , Xueming Fu , Junfeng Jiang , Qiang Zeng , Ye Tang , Zhengming Chen , Luming Nong , Feng Wang
show 1 more author
S. Kevin Zhou
This is my paper
classification cs.CVcs.AI
keywords CT/X-rayregistration2D/3DRadiativeGaussians3DGaussianSplattingbiplanarX-rayreconstructionvertebral-levelcounterfactualattentionimage-guidednavigation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tries to show that the hard 2D/3D problem of matching intraoperative X-rays to a preoperative spine CT becomes easier if you lift it into 3D first: reconstruct a volumetric Radiative Gaussians (RadGS) model from just two X-rays, then register that volume to the CT volume with a learned 6-DOF pose network. It argues that this joint reconstruction-plus-registration framework beats both 'render and compare' iterative methods and earlier reconstruction-based approaches, with reported mean target registration error of 1.14 mm, success rate 93.33% under 2 mm error, and runtime 0.82 s on an in-house dataset. The authors also claim the approach is model-agnostic across backbone networks and benefits from a three-stage pre-training that moves from simulated DRRs to real X-rays to patient-specific data. If true, this would give image-guided spine navigation an accurate, near-real-time way to align a pre-op CT to two live X-rays without iterative DRR rendering.

What carries the argument

The carrying mechanism is 'Radiative Gaussians' (RadGS), a 3D Gaussian-splatting representation in which each Gaussian carries a central density, position, and covariance; an X-ray rasterizer projects these Gaussians to form a DRR-like image, and a density voxelizer turns them into a volume for registration. The counterfactual attention module quantifies which image regions actually drive the reconstruction by comparing factual predictions with predictions made under randomly intervened attention maps, forcing the encoder to focus on vertebra bodies rather than adjacent structures. The registration module is a straightforward image-encoder plus pose-head that maps the six-DOF transform from

What would settle it

Run the method on a public multi-center spine dataset with independent ground truth, e.g., pose from an optically tracked C-arm, using vertebrae not seen in training and withholding the patient-specific third pre-training stage; if the success rate under 2 mm falls to the level of the compared methods, then the reported 93.33% depends on per-patient adaptation rather than the joint reconstruction-plus-registration mechanism itself.

Watch

Extended reading notes

Core claim

The central claim is that biplanar X-rays contain enough information to reconstruct a Radiative Gaussians (RadGS) volume of a vertebra, and that this volume—not the original X-rays—should be the object registered to the CT. The reconstruction module encodes the two X-rays, uses counterfactual attention (CAL) to suppress interference from overlapping vertebrae, and outputs Gaussians (density, position, covariance) which are rendered back to X-rays and voxelized into a volume. The registration module concatenates this voxelized RadGS volume with the segmented CT volume and regresses the rigid 6-DOF transform. The two modules are first trained separately, then jointly so that registration error

Load-bearing premise

That a vertebra's true 3D density and shape can be recovered well enough from only two X-rays—through the learned reconstruction—that a network finding the pose of that reconstructed volume is solving the same problem as aligning the patient; if two-view reconstruction is ambiguous or biased in depth, the reported sub-2 mm errors do not necessarily transfer to real alignment.

Editorial extensions

If this is right

  • Vertebral CT/X-ray registration no longer needs to iterate render-and-compare during inference; the 0.82 s runtime means the alignment can be completed within a typical intraoperative pause.
  • Because the reconstruction is optimization-free and uses only two X-rays, the method could work with standard AP/LA C-arm setups rather than dense-view sweeps.
  • Joint training ties reconstruction quality to registration accuracy, so improvements in either module should translate into lower target registration error.
  • The model-agnostic backbone results suggest the pipeline can be re-hosted on different network architectures without redesigning the training scheme.
  • The three-stage pre-training path suggests simulated data alone is insufficient; real and patient-specific data are needed for clinical-grade accuracy.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A natural next test is whole-spine or multi-level registration: the current evaluation is vertebral-level, so the method's sensitivity to inter-vertebral pose differences remains open.
  • Because the third pre-training stage uses the target CT with DRRs, the method implicitly assumes a per-patient calibration step before surgery; a reader might infer that removing this stage would lower accuracy, as the ablation results suggest.
  • If the reconstructed RadGS volume faithfully encodes geometry, the same volume could be reused for tasks other than registration—e.g., pre-operative planning or automatic verification of C-arm positioning—without extra DRR generation.
  • A direct extension would be replacing the rigid pose head with a deformation field to handle soft tissue and non-rigid regions, which the paper lists as future work.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper presents RadGS-Reg, a learning-based framework for vertebral-level CT/X-ray registration. The method first reconstructs a 3D Radiative Gaussians (RadGS) volume from biplanar AP/LA X-rays using an image encoder, a counterfactual attention module, an X-ray rasterizer, and a density voxelizer. This reconstructed volume is then registered to a preoperative CT volume by a 3D/3D registration network that regresses a 6-DOF rigid pose. Training is performed in stages: (1) simulated DRRs from the public VERSE dataset, (2) real intraoperative X-rays from an in-house 10-case dataset, and (3) synthetic DRRs and CT pairs generated from the target CT volume of each cross-validation fold. On the in-house dataset the authors report an SSIM of 94.51%, PSNR of 28.80 dB, mTRE of 1.14 mm, success rate of 93.33%, and runtime of 0.82 s, and they claim state-of-the-art performance over existing reconstruction and registration methods.

Significance. If the reported results hold under a clean evaluation protocol, the paper makes a useful contribution: it recasts 2D/3D registration as 3D/3D registration via a learned radiologic Gaussian representation, introduces a counterfactual attention mechanism to handle noisy biplanar X-rays, and demonstrates that joint/sequential training of reconstruction and registration can improve both. The availability of code and the internal consistency of the ablation study are strengths. However, the central significance claim is currently undermined by the evaluation protocol, because the third pre-training stage uses the held-out fold's target CT volume to generate additional training data before inference. This makes the comparison with existing methods unfair and means the reported accuracy may reflect target-specific adaptation rather than generalizable registration accuracy. The small in-house dataset (10 cases, 30 vertebral-level samples) further limits the strength of the conclusions.

major comments (3)
  1. [Sec. 3.2 / Table 2] The three-stage pre-training includes a stage trained on the target CT volume of the held-out fold: 'For each fold, the validation set of preoperative CT volumes generated 1,800 biplanar DRR pairs and 600 CT pairs for the third stage.' Since the validation set is the test fold, the registration and reconstruction modules are fine-tuned on the exact anatomy they are later asked to register. This is transductive rather than circular (the test X-rays are not used), but it gives RadGS-Reg a patient-specific shape and density prior that the compared baselines do not receive. Table 2 shows that this stage is responsible for a large gain: SR rises from 80.00% (P2) to 93.33% (PFULL) and mTRE falls from 1.36 mm to 1.14 mm. The authors must either re-run the evaluation without the target-CT third stage, or clearly reframe the method as a patient-specific adaptation pipeline and give the baselines
  2. [Sec. 3.1 / Table 1] The in-house evaluation is very small: 10 intraoperative cases and 30 vertebral-level samples, with five-fold cross-validation. Multiple vertebrae from the same patient may be correlated, yet the statistical significance test ('*', p<0.05) is reported without stating the test procedure, whether the samples are treated as independent, or whether correction for multiple comparisons was applied. The paper should report per-patient or per-vertebra error distributions, confidence intervals, and ideally an independent external dataset. Without this, the 1.14 mm mTRE and 93.33% SR are not sufficient evidence that the method generalizes beyond this particular acquisition setup.
  3. [Sec. 3.3 / Table 1] The comparison with DiffPose, DDGS-CT, and TS-SAR does not specify the initialization or pose perturbation range used for these iterative baselines. The proposed method directly regresses the pose, whereas render-and-compare methods are initialization-dependent. If the baselines were initialized far from the true pose or were not given the benefit of a multi-start search, their low success rates may be an artifact of the protocol. The authors should detail the common evaluation conditions, including initial pose distributions, number of restarts, and whether the same target-CT adaptation was applied to any baseline.
minor comments (5)
  1. [Eq. (1)] The notation is confusing: the text says 'replacing the factual attention maps¯A with random weights,' but the equation uses A and a counterfactual distribution γ. Please define A, A-bar, and the intervention do(·) precisely.
  2. [Sec. 2, Eq. (2)] The text says the reconstruction comparison uses 'Lrec from Eq. (2)', but Eq. (2) is named LCAL-rec and Lrec is only mentioned in the following sentence. Please define Lrec explicitly and state which loss is used for each compared reconstruction method.
  3. [Sec. 3.1] Typo: 'real-scent intraoperative cases' should be 'real-scene' or 'real clinical intraoperative cases'.
  4. [Sec. 3.2 / Table 1] The hardware description reads 'AMD Core EYPC 7R32'; the correct vendor spelling is 'AMD EPYC'. Also, the table caption mentions 'results marked by the dotted line,' but no dotted line appears in the table; please align the text with the table.
  5. [Sec. 3.2] The definition of CR is unclear: 'the acceptable CR interval was set to 5 mm' and Table 1 reports CR as ranges such as '20-25' and '10-15'. Please define capture range precisely and explain how the interval is computed from the reported tests.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity; target-CT pre-training is a generalization concern, not a by-construction equivalence.

full rationale

RadGS-Reg's core chain—biplanar X-rays → RadGS reconstruction → 3D/3D registration against CT—is not an algebraic consequence of its inputs. Ground-truth poses were recorded externally from C-arm acquisition, and at test time the network must infer a pose from real X-rays that were not used as training images. The pose output is therefore non-trivially determined by the X-ray content, not by construction. The only notable protocol choice is the third pre-training stage (Sec. 3.2): for each five-fold split, the held-out fold's CT volumes are used to synthesize 1,800 biplanar DRR pairs and 600 CT pairs for fine-tuning. This means the exact CT used as the registration target is also used to train the model before evaluation. That is a form of transductive leakage and explains part of the gain in Table 2 (SR 80.00% → 93.33% from P2 to Pfull), but it does not make the reported mTRE circular: the synthetic training poses are randomly sampled, the real X-ray is not in the training set, and the network must still extract the actual pose from the X-ray. The paper also cites a co-author's work [14] for X-ray vertebral-level detection, but that is a preprocessing tool, not a load-bearing mathematical premise. No equation reduces to its own input, and no uniqueness claim is imported from self-citation. The central derivation is self-contained; the evaluation protocol's leakage is a correctness/external-validity issue, not a circularity issue.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The method leans on several modeling assumptions: the two-view reconstruction problem is well-posed, the Gaussian splatting renderer faithfully simulates X-rays, and the recorded GT poses are error-free. The loss weights are taken from prior work rather than tuned on this data.

free parameters (4)
  • lambda1 = 0.2
    Weight for SSIM loss in Eq. (2), set as per R2-GS [26].
  • lambda2 = 0.05
    Weight for TV loss in Eq. (2), set as per R2-GS [26].
  • lambda = 0.02
    Weight for geodesic loss in Eq. (3), set as suggested by [9].
  • Training schedule = 300 epochs, lr 0.1 decay 0.5 per 50 epochs, batch size 8
    Optimization hyperparameters; not justified by a search, chosen by hand.
assumptions (5)
  • domain assumption Two biplanar X-rays with known geometry are sufficient to reconstruct a vertebra's 3D shape given learned prior.
    Central to RecM; enables reconstruction from only two views in Sec. 2.
  • domain assumption Radiative Gaussian splatting [26] accurately models X-ray image formation and CT density projection.
    Relied on for the reconstruction loss Eq. (2) and the whole pipeline.
  • domain assumption The voxelized RadGS volume is a valid intermediate representation for 3D/3D registration against CT.
    Used to concatenate reconstructed volume with CT in RegM (Sec. 2).
  • domain assumption The GT poses recorded during C-arm circular acquisitions are accurate.
    Used as supervision in Eq. (3) and for evaluation.
  • domain assumption Counterfactual attention learning [22] improves attention quality in this X-ray setting.
    Motivates the CAL module in Sec. 2.

how reviews work

0 comments
Cite this review

Pith. "Pith review of RadGS-Reg: Registering Spine CT with Biplanar X-rays via Joint 3D Radiative Gaussians Reconstruction and 3D/3D Registration." pith.science (2026). https://pith.science/paper/Z6HIMFNI

@misc{pith2026250821154,
  author       = {Pith},
  title        = {Pith review of: RadGS-Reg: Registering Spine CT with Biplanar X-rays via Joint 3D Radiative Gaussians Reconstruction and 3D/3D Registration},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Z6HIMFNI}},
  note         = {Machine review of arXiv:2508.21154}
}
read the original abstract

Computed Tomography (CT)/X-ray registration in image-guided navigation remains challenging because of its stringent requirements for high accuracy and real-time performance. Traditional "render and compare" methods, relying on iterative projection and comparison, suffer from spatial information loss and domain gap. 3D reconstruction from biplanar X-rays supplements spatial and shape information for 2D/3D registration, but current methods are limited by dense-view requirements and struggles with noisy X-rays. To address these limitations, we introduce RadGS-Reg, a novel framework for vertebral-level CT/X-ray registration through joint 3D Radiative Gaussians (RadGS) reconstruction and 3D/3D registration. Specifically, our biplanar X-rays vertebral RadGS reconstruction module explores learning-based RadGS reconstruction method with a Counterfactual Attention Learning (CAL) mechanism, focusing on vertebral regions in noisy X-rays. Additionally, a patient-specific pre-training strategy progressively adapts the RadGS-Reg from simulated to real data while simultaneously learning vertebral shape prior knowledge. Experiments on in-house datasets demonstrate the state-of-the-art performance for both tasks, surpassing existing methods. The code is available at: https://github.com/shenao1995/RadGS_Reg.

Figures

Figures reproduced from arXiv: 2508.21154 by the authors.

Figure 1
Figure 1. The overall framework of RadGS-Reg (a-d) and the biplanar X-rays RadGS reconstruction module (e). The framework consists of four parts: (a) 3D reconstruction, (b) 3D/3D registration, (c) synergistic training, and (d) inference phase. training efficiency through module-aware and data-aware strategies, with results surpassing state-of-the-art methods in 3D reconstruction and registration. 2 Method The proposed end-to-… view at source ↗
Figure 2
Figure 2. Qualitative comparison with the different reconstruction methods. For each case separated by a dashed line, column 1 displays attention maps from RecM without CAL, while column 2 shows those with CAL, with the color bar indicating the voxel intensity distribution. Ablation Study Ablation studies were conducted to assess the efficacy of the modules within RadGS-Reg, encompassing the CAL module, synergistic train￾ing,… view at source ↗
Figure 3
Figure 3. Qualitative comparison with different registration methods. Each row represents an individual case [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

28 extracted references · 20 canonical work pages

  1. [1]

    Medical Image Analysis73, 102166 (2021)

    Verse: A vertebrae labelling and segmentation benchmark for multi-detector ct images. Medical Image Analysis73, 102166 (2021)

    work page 2021

  2. [2]

    IEEE Transactions on Circuits and Systems for Video Technology (2025)

    Bao, Y., Ding, T., Huo, J., Liu, Y., Li, Y., Li, W., Gao, Y., Luo, J.: 3d gaussian splatting: Survey, technologies, challenges, and opportunities. IEEE Transactions on Circuits and Systems for Video Technology (2025)

    work page 2025

  3. [3]

    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)

    work page 2024

  4. [4]

    In: European Conference on Computer Vision

    De Roovere, P., Daems, R., Croenen, J., Bourgana, T., de Hoog, J., Wyffels, F.: Cendernet: Center and curvature representations for render-and-compare 6d pose estimation. In: European Conference on Computer Vision. pp. 97–111. Springer (2022)

    work page 2022

  5. [5]

    arXiv preprint arXiv:2010.11929 (2020)

    Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al.: An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929 (2020)

    arXiv 2010

  6. [6]

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

    Fu, X., Li, Y., Tang, F., Li, J., Zhao, M., Teng, G.J., Zhou, S.K.: 3dgr-car: Coro- nary artery reconstruction from ultra-sparse 2d x-ray views with a 3d gaussians representation. In: International Conference on Medical Image Computing and Computer-Assisted Intervention. pp. 14–24. Springer (2024)

    work page 2024

  7. [7]

    IEEE transactions on medical imaging (2023) 10 A

    Gao, C., Feng, A., Liu, X., Taylor, R.H., Armand, M., Unberath, M.: A fully differ- entiable framework for 2d/3d registration and the projective spatial transformers. IEEE transactions on medical imaging (2023) 10 A. Shen et al

    work page 2023

  8. [8]

    arXiv preprint arXiv:2406.02518 (2024)

    Gao, Z., Planche, B., Zheng, M., Chen, X., Chen, T., Wu, Z.: Ddgs-ct: Direction- disentangled gaussian splatting for realistic volume rendering. arXiv preprint arXiv:2406.02518 (2024)

    arXiv 2024

Show all 28 references
  1. [9]

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

    Gopalakrishnan, V., Dey, N., Golland, P.: Intraoperative 2d/3d image registration via differentiable x-ray rendering. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 11662–11672 (2024)

    2024

  2. [10]

    Interna- tional journal of computer assisted radiology and surgery15, 759–769 (2020)

    Grupp, R.B., Unberath, M., Gao, C., Hegeman, R.A., Murphy, R.J., Alexander, C.P.,Otake,Y.,McArthur,B.A.,Armand,M.,Taylor,R.H.:Automaticannotation of hip anatomy in fluoroscopy for robust and efficient 2d/3d registration. Interna- tional journal of computer assisted radiology a...

    2020

  3. [11]

    He,K.,Zhang,X.,Ren,S.,Sun,J.:Deepresiduallearningforimagerecognition.In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 770–778 (2016)

    2016

  4. [12]

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

    Huang, G., Liu, Z., Van Der Maaten, L., Weinberger, K.Q.: Densely connected convolutional networks. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 4700–4708 (2017)

    2017

  5. [13]

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

    Jaganathan, S., Kukla, M., Wang, J., Shetty, K., Maier, A.: Self-supervised 2d/3d registration for x-ray to ct image fusion. In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. pp. 2788–2798 (2023)

    2023

  6. [14]

    In: 2024 IEEE Inter- national Conference on Bioinformatics and Biomedicine (BIBM)

    Jiang, J., Shuai, L., Zeng, Q.: Ablspinelevelcheck: Localization of vertebral lev- els on fluoroscopy via semi-supervised abductive learning. In: 2024 IEEE Inter- national Conference on Bioinformatics and Biomedicine (BIBM). pp. 3347–3350. IEEE (2024)

    2024

  7. [15]

    ACM Trans

    Kerbl, B., Kopanas, G., Leimkühler, T., Drettakis, G.: 3d gaussian splatting for real-time radiance field rendering. ACM Trans. Graph.42(4), 139–1 (2023)

    2023

  8. [16]

    arXiv preprint arXiv:2212.06870 (2022)

    Labbé, Y., Manuelli, L., Mousavian, A., Tyree, S., Birchfield, S., Tremblay, J., Carpentier, J., Aubry, M., Fox, D., Sivic, J.: Megapose: 6d pose estimation of novel objects via render & compare. arXiv preprint arXiv:2212.06870 (2022)

    2022 arXiv

  9. [17]

    Medical & Biological Engineering & Computing60(8), 2271–2289 (2022)

    Naik, R.R., Anitha, H., Bhat, S.N., Ampar, N., Kundangar, R.: Realistic c-arm to pct registration for vertebral localization in spine surgery: A hybrid 3d-2d registra- tion framework for intraoperative vertebral pose estimation. Medical & Biological Engineering & Computing60(8...

    2022

  10. [18]

    IEEE transactions on medical imaging31(4), 948–962 (2011)

    Otake, Y., Armand, M., Armiger, R.S., Kutzer, M.D., Basafa, E., Kazanzides, P., Taylor, R.H.: Intraoperative image-based multiview 2d/3d registration for image- guided orthopaedic surgery: incorporation of fiducial-based c-arm tracking and gpu-acceleration. IEEE transactions o...

    2011

  11. [19]

    In: VISIGRAPP (5: VISAPP)

    Payer, C., Stern, D., Bischof, H., Urschler, M.: Coarse to fine vertebrae localization and segmentation with spatialconfiguration-net and u-net. In: VISIGRAPP (5: VISAPP). pp. 124–133 (2020)

    2020

  12. [20]

    Penney, G.P.,Weese,J.,Little, J.A.,Desmedt,P.,Hill,D.L., etal.:Acomparison of similaritymeasuresforusein2-d-3-dmedicalimageregistration.IEEEtransactions on medical imaging17(4), 586–595 (1998)

    1998

  13. [21]

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

    Ponimatkin, G., Labbé, Y., Russell, B., Aubry, M., Sivic, J.: Focal length and object pose estimation via render and compare. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 3825–3834 (2022)

    2022

  14. [22]

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

    Rao, Y., Chen, G., Lu, J., Zhou, J.: Counterfactual attention learning for fine-grained visual categorization and re-identification. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 1025–1034 (2021)

    2021

  15. [23]

    IEEE transactions on medical imaging38(2), 470–481 (2018) RadGS-Reg 11

    Salehi,S.S.M.,Khan,S.,Erdogmus,D.,Gholipour,A.:Real-timedeepposeestima- tion with geodesic loss for image-to-template rigid registration. IEEE transactions on medical imaging38(2), 470–481 (2018) RadGS-Reg 11

    2018

  16. [24]

    SIAM Journal on Imaging Sciences1(3), 248–272 (2008)

    Wang, Y., Yang, J., Yin, W., Zhang, Y.: A new alternating minimization algorithm for total variation image reconstruction. SIAM Journal on Imaging Sciences1(3), 248–272 (2008)

    2008

  17. [25]

    IEEE transactions on image processing 13(4), 600–612 (2004)

    Wang, Z., Bovik, A.C., Sheikh, H.R., Simoncelli, E.P.: Image quality assessment: from error visibility to structural similarity. IEEE transactions on image processing 13(4), 600–612 (2004)

    2004

  18. [26]

    arXiv preprint arXiv:2405.20693 (2024)

    Zha, R., Lin, T.J., Cai, Y., Cao, J., Zhang, Y., Li, H.: R 2-gaussian: Rectify- ing radiative gaussian splatting for tomographic reconstruction. arXiv preprint arXiv:2405.20693 (2024)

    2024 arXiv

  19. [27]

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

    Zhang, B., Faghihroohi, S., Azampour, M.F., Liu, S., Ghotbi, R., Schunkert, H., Navab, N.: A patient-specific self-supervised model for automatic x-ray/ct regis- tration. In: International Conference on Medical Image Computing and Computer- Assisted Intervention. pp. 515–524. ...

    2023

  20. [28]

    Biomedical Signal Processing and Control 95, 106384 (2024)

    Zhao, H., Zhu, W., Deng, X., Zhang, G., Zou, W., et al.: Automatic 2d/3d spine registration based on two-step transformer with semantic attention and adaptive multi-dimensional loss function. Biomedical Signal Processing and Control 95, 106384 (2024)

    2024

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.