NeRF-based CBCT Reconstruction needs Normalization and Initialization

Dai Sun; Han Li; Nassir Navab; Qingpeng Kong; S. Kevin Zhou; Yujia Li; Zhicheng Li; Zhiwei Cheng; Zhuowei Xu

REVIEW 3 major objections 6 minor 1 cited by

NeRF-based CBCT Reconstruction needs Normalization and Initialization

T0 review · 3 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Layer normalization and a one-case pretrained initialization fix the core training mismatch in NeRF-based CBCT reconstruction.

desk verdict The local-global mismatch framing is useful, but the gains are attributed to LN+MCI while an unablated FFT-masking step is folded into the evaluated method; the central claim needs an ablation rerun. read the letter →

arxiv 2506.19742 v1 pith:LVX3R2LC submitted 2025-06-24 eess.IV cs.AIcs.CV

Zhuowei Xu , Han Li , Dai Sun , Zhicheng Li , Yujia Li , Qingpeng Kong , Zhiwei Cheng , Nassir Navab

show 1 more author

S. Kevin Zhou

This is my paper · ORCID

classification eess.IVcs.AIcs.CV

keywords CBCTreconstructionNeRFhashencodinglayernormalizationinitializationsparse-viewCT3DGaussiansplattingtransferlearning

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

The paper argues that NeRF-based cone-beam CT (CBCT) reconstruction suffers from a local-global optimization mismatch: each training step updates only a small subset of the hash encoder's parameters but all of the neural network's parameters, so the features the network sees keep shifting underneath it. To fix this, the authors insert a Layer Normalization between the hash encoder and the network, and initialize the network with weights pretrained on the dense volume of one abdomen case, calling this Mapping Consistency Initialization. They report that this few-line change makes training converge faster and raises reconstruction quality above the NAF baseline, reaching quality comparable to or better than 3D Gaussian Splatting methods while keeping a similar training time. Experiments cover 128 CT cases from four datasets and seven anatomical regions. The central claim is that normalization plus one-case initialization is a general cure for the mismatch in hash-NeRF sparse-view reconstruction.

What carries the argument

The central mechanism is the pairing of a Normalized Hash Encoder with Mapping Consistency Initialization. The former is a Layer Normalization layer placed between the multiresolution hash grid and the MLP; it forces hash features from different subsets of the grid to share one mean and variance, so the MLP no longer sees a moving target. The latter bypasses rendering: a full NeRF is trained directly on the voxel attenuation values of one abdomen case with an $\ell^1$ loss, and only its LN and MLP weights are copied into the reconstruction model. FFT-based channel masking of noisy hash channels is a secondary mechanism. The work of these components is to stabilize the early training dynamics that the paper identifies as the bottleneck.

What would settle it

Train NAF+LN+MCI with the MCI weights pretrained on a case sampled from a scanner or dataset entirely outside the four evaluation datasets, and compare against random initialization; if the quality gap mostly disappears, the transfer claim fails. Independently, record a fixed batch of hash features, run them through the MLP at two different training times, and measure the L1 drift in outputs: if adding Layer Normalization does not reduce that drift, the proposed mismatch mechanism is not the active cause.

Watch

Extended reading notes

Core claim

The paper's central discovery is that unstable training in hash-NeRF-based CBCT reconstruction is caused by a local-global optimization mismatch, and that two cheap interventions substantially reduce it: Layer Normalization on the hash features, and initialization of the MLP from a dense-volume pretraining on a single ground-truth case. The normalized hash encoder keeps the features' global mean and variance stable across training steps, and the Mapping Consistency Initialization gives the network a sensible global mapping from coordinates to attenuation values before sparse-view training begins. In the paper's experiments, the resulting method (NAF+LN+MCI) outperforms NAF on every dataset, outperforms R2_Gaussian on most datasets, and reaches convergence in about 18 minutes, matching the speed of 3DGS.

Load-bearing premise

The gains rest on the premise that an MLP pretrained on the ground-truth attenuation values of a single abdomen case transfers to every other case and body region, and that the improvement comes from the initialization itself rather than from the pretraining case sharing its distribution with the test datasets.

Editorial extensions

If this is right

Under the same 50-view protocol, NAF+LN+MCI improves PSNR and SSIM over NAF on the chest, abdomen, and head datasets, and surpasses R2_Gaussian on most of the four R2_Gaussian cases.
The method reaches convergence in about 18 minutes, matching 3DGS speed while exceeding its reconstruction quality on most datasets.
Better Average Segment Dice scores than R2-Gaussian indicate the reconstructions preserve anatomical structure, not just pixel-level metrics.
The recipe requires only a few lines of code and no paired data at reconstruction time, so it can be dropped into existing hash-NeRF CBCT pipelines.
Ablation shows each component, Layer Normalization and Mapping Consistency Initialization, independently accelerates convergence and improves final quality.
The improvements hold across 128 cases spanning seven anatomical regions, suggesting the recipe is not tuned to one body part.

Reading between the lines

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

The pretraining is a pure coordinate-to-attenuation regression, so the same MCI weights should transfer to other forward models, such as different detector geometries or noise models, as long as the coordinate frame is normalized; the paper does not test this.
The paper's mechanism predicts that Layer Normalization alone should reduce the output drift of the MLP on fixed hash features in any hash-encoded scene regression, which could be verified outside CBCT, for instance in instant-NGP view synthesis.
The FFT channel-masking result suggests some hash channels carry dataset-specific noise; one could test whether the masked channels identified in the abdomen pretraining case generalize to chest and head, or whether each region needs its own mask.
If the transfer holds, this is effectively a pretraining recipe for inverse problems, and the one-case requirement could be stress-tested by measuring how reconstruction quality degrades as the pretraining case moves further in feature space from the target anatomy.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Colleague,

Worth a look if you follow hash-NeRF methods in sparse-view CT. The paper's framing—that the hash encoder's local sparse updates and the MLP's global dense updates create a train-time mismatch—is a genuinely useful way to talk about why hash-NeRF training is unstable. Adding LayerNorm between the hash grid and the MLP, plus initializing the MLP from a pretrained dense-volume model, speeds up NAF convergence and lifts reconstruction quality on their test bed. The gains look real, and the experiments cover 128 cases across four datasets, with public code.

But the paper's central attribution is not yet proven. The stress-test note is accurate: Sec. 2.3 introduces an FFT-based channel-masking step that is part of the evaluated method but is absent from the title, the abstract, the method name (NAF+LN+MCI), and the ablation in Fig. 8. If the final model includes FFT masking, the comparison in Table 1 conflates three interventions, not two. You cannot tell whether LN, MCI, or the masking drives the improvement. This is not a footnote; it is the load-bearing claim.

Second, MCI pretraining uses a ground-truth CT volume of one abdomen case. The paper even says the pretraining 'uses ground truth information' (Sec. 2.1) and the pretraining case comes from Pancreas_CT, which is also one of the test datasets. That is not the same as test-case leakage, but it is a supervised pretraining signal from the evaluation distribution, and no baseline is granted an equivalent. A cleaner baseline would give all methods the same initialization, or show that a random-case pretraining produces the same gains.

Minor but not negligible: Table 1 reports single numbers with no error bars or significance tests, and some deltas over NAF are small (e.g., head PSNR 34.53 vs 34.09).

What the paper does well: it names a real training dynamic, tests a simple recipe across several anatomies, and reports segmentation-based evaluation in addition to PSNR/SSIM. The implementation is public.

Bottom line: the hypothesis is plausible and the authors did the empirical work, but the reported experiments do not isolate the claimed mechanisms. I would still send this to peer review—the question is interesting and the paper is substantial enough to warrant referee time—but I would ask for an ablation that separates FFT masking, error bars, and a fairer initialization comparison before the central claim can be accepted.

Referee Report

3 major / 6 minor

Summary. This paper addresses sparse-view CBCT reconstruction with hash-encoded NeRF. It attributes training instability to a local-global optimization mismatch between the hash grid and the MLP, and proposes two remedies: a Layer Normalization layer between hash features and the MLP, and a Mapping Consistency Initialization (MCI) that pretrains the MLP on a single ground-truth abdomen volume using a voxel-to-voxel loss. The method is evaluated against FDK, SART, NAF, SAX-NeRF, and R2-Gaussian on 128 CT cases from four datasets spanning seven anatomical regions, with PSNR/SSIM and segmentation Dice. The paper reports that NAF+LN+MCI converges faster and achieves higher image quality than NAF, and surpasses R2-Gaussian on most datasets with comparable runtime. The authors claim that normalization and initialization are sufficient to alleviate the local-global mismatch.

Significance. The proposed recipe is simple, easy to implement, and accompanied by a public code repository, which are genuine strengths. The evaluation is broad: 128 cases, four datasets, seven anatomical regions, and a clinical segmentation metric. If the attribution to LN and MCI were established, this would be a practically useful and general improvement for hash-NeRF sparse-view CBCT reconstruction. However, the current manuscript does not isolate the two claimed components because Section 2.3 introduces an additional FFT-based channel-masking step that is absent from the ablation and abstract, and the main comparison lacks statistical uncertainty quantification and leaves the pretraining-data overlap with the test distribution unclear. These issues prevent the central causal claim from being accepted as stated.

major comments (3)

[Sec. 2.3 / Fig. 8 / Table 1] The paper introduces a third intervention that the attribution ignores. The text states that at the channel level of the hash features, FFT is used to distinguish and mask noisy channels during training to improve performance, but the abstract, the method name NAF+LN+MCI, and the ablation in Fig. 8 only mention LN and MCI. If the final model in Table 1 includes FFT masking, the comparison to NAF conflates three changes rather than two, and the claim that normalization and initialization alone produce the reported gains is not supported. Please add ablation arms that isolate FFT masking (e.g., NAF+LN, NAF+LN+FFT, NAF+LN+MCI, NAF+LN+MCI+FFT), report the corresponding Table 1 numbers, and either include FFT masking in the method name and abstract or remove it from the method description.
[Table 1 / Fig. 6] The main quantitative comparison reports point estimates without error bars, confidence intervals, or significance tests. Several differences are small relative to likely run-to-run variability (e.g., Head PSNR 34.53 vs 34.09 for NAF, and GS_foot PSNR 31.64 vs 31.3 for NAF), and the conclusion that the method surpasses 3DGS depends on these means. Please report means and standard deviations over multiple runs or over the individual cases in each dataset, and perform paired significance tests for the headline comparisons.
[Sec. 2.4 / Sec. 3.1 / Table 1] MCI pretraining uses a ground-truth attenuation volume of one abdomen case while the baselines receive no such pretraining, and the manuscript does not specify whether this case belongs to the 82 Pancreas_CT abdomen cases used in evaluation or is excluded from all test sets. If the pretraining case is drawn from the test distribution, the comparison leaks information; even if it is held out, the choice of an abdomen case from the same public dataset as the abdomen test set needs a transferability control. Please state the source of the pretraining case, verify that it is not in any test set, and report results when the pretraining volume is taken from a different anatomical region.

minor comments (6)

[Sec. 1] The word 'Sepcifically' should be 'Specifically'.
[Fig. 1] The caption contains the typo 'Hash Gird' instead of 'Hash Grid'.
[Fig. 8] The ablation figure reports SSIM curves for three images but does not state which cases or anatomical regions are used, nor does it provide the final SSIM values; please add this information.
[Throughout] The paper uses the names R2_Gaussian and R2-Gaussian interchangeably; please unify the notation.
[Sec. 3.1] The implementation details omit the hash-grid configuration (number of levels, feature dimension, resolution) and MLP architecture; providing these is important because the hash-grid size affects the local-global mismatch.
[Fig. 6 / Sec. 3.2] The segment Dice scores are shown only graphically; please include the numerical values in the text or a table. Also, the phrase 'on three cases7' in Sec. 3.2 contains an unresolved footnote marker, and Fig. 7 is not explicitly discussed; please fix the reference.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the improvements are empirically tested against external baselines, and no prediction is forced by construction or by self-citation.

full rationale

This paper makes no analytic derivation whose conclusion is equivalent to its premises. The central claim is that adding Layer Normalization and a Mapping Consistency Initialization to NAF improves sparse-view CBCT reconstruction. That claim is evaluated by comparing NAF+LN+MCI against FDK, SART, NAF, SAX-NeRF, and R2-Gaussian on public datasets with PSNR, SSIM, and segmentation Dice. LN is a standard normalization layer inserted between the hash encoder and MLP, and MCI initializes the MLP by voxel-wise supervision on one ground-truth abdomen case; neither component is defined in terms of the reported reconstruction metrics, and neither is fitted to the test-set outputs. The MCI pretraining uses ground truth from the same public data distribution as the evaluation, which is a legitimate data-leakage / generalization concern, but it is not circular reasoning because the pretraining objective (Eq. 2) does not encode the test projections or the metrics being predicted. The paper also describes an FFT-based channel masking step in Sec. 2.3 that is not included in the abstract, method name, or ablation in Fig. 8; this is an attribution gap and a correctness risk, but it is not a case where a prediction reduces to its inputs by construction. There are no self-citations used as load-bearing arguments, no imported uniqueness theorems, and no renaming of a known result as new organization. Therefore the circularity score is 0.

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

No new physical entities or mathematical constants are introduced. The method depends on empirical assumptions about transferability of pretraining, the validity of the Beer-Lambert forward model, and the hypothesized mismatch mechanism; these are listed as axioms.

free parameters (4)

MCI pretraining source case = one abdomen case from Pancreas_CT
The choice of a single abdomen case for dense-volume pretraining is arbitrary; no sensitivity analysis is reported, so the result may depend on which case is selected.
Number of sparse views = 50 views
All main experiments fix 50 simulated projections; the method is evaluated at other view counts only in Fig. 7, not in the main quality table.
Training epochs = 3000 (NeRF-like), 30000 (R2_Gaussian)
Epoch counts are chosen to ensure convergence; no early stopping or convergence criterion is reported, and pretraining duration for MCI is not given.
FFT hash-channel mask = unspecified
The paper says FFT is used to distinguish and mask noisy channels (Sec 2.3) without giving the masking criterion; this hand-designed choice can affect reconstruction quality and is not ablated.

assumptions (4)

domain assumption X-ray intensity attenuation follows the Beer-Lambert law, so projection values equal line integrals of attenuation coefficients.
Used in Eq. (1) for both rendering and loss; assumes a monoenergetic forward model with no scatter or beam hardening.
ad hoc to paper The local-global optimization mismatch is the cause of slow convergence and degraded quality in hash-NeRF.
The paper presents indirect evidence (Fig. 3, 4) but no formal proof or controlled experiment isolating this mechanism.
domain assumption Dense-volume pretraining on one ground-truth abdomen volume transfers to all other cases and anatomical regions.
The whole MCI strategy rests on this transferability; only one pretraining case is used with no sensitivity analysis.
ad hoc to paper Layer Normalization equalizes hash features without discarding information needed for attenuation prediction.
The paper assumes LN only aligns feature distributions and does not harm discriminative content; no information-theoretic or ablative evidence is provided beyond overall metrics.

how reviews work

0 comments

Cite this review

Pith. "Pith review of NeRF-based CBCT Reconstruction needs Normalization and Initialization." pith.science (2026). https://pith.science/paper/LVX3R2LC

@misc{pith2026250619742,
  author       = {Pith},
  title        = {Pith review of: NeRF-based CBCT Reconstruction needs Normalization and Initialization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LVX3R2LC}},
  note         = {Machine review of arXiv:2506.19742}
}

read the original abstract

Cone Beam Computed Tomography (CBCT) is widely used in medical imaging. However, the limited number and intensity of X-ray projections make reconstruction an ill-posed problem with severe artifacts. NeRF-based methods have achieved great success in this task. However, they suffer from a local-global training mismatch between their two key components: the hash encoder and the neural network. Specifically, in each training step, only a subset of the hash encoder's parameters is used (local sparse), whereas all parameters in the neural network participate (global dense). Consequently, hash features generated in each step are highly misaligned, as they come from different subsets of the hash encoder. These misalignments from different training steps are then fed into the neural network, causing repeated inconsistent global updates in training, which leads to unstable training, slower convergence, and degraded reconstruction quality. Aiming to alleviate the impact of this local-global optimization mismatch, we introduce a Normalized Hash Encoder, which enhances feature consistency and mitigates the mismatch. Additionally, we propose a Mapping Consistency Initialization(MCI) strategy that initializes the neural network before training by leveraging the global mapping property from a well-trained model. The initialized neural network exhibits improved stability during early training, enabling faster convergence and enhanced reconstruction performance. Our method is simple yet effective, requiring only a few lines of code while substantially improving training efficiency on 128 CT cases collected from 4 different datasets, covering 7 distinct anatomical regions.

Figures

Figures reproduced from arXiv: 2506.19742 by the authors.

**Figure 2.** Our pipeline of pre-training and normal reconstruction. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 4.** Feature PCA results of abdomen, chest and head. the neural network converges slowly and exhibits significant confusion; incorporating LN enables the network to converge quickly and stably. We place LN between the neural network and the hash encoder. Furthermore, at the channel level of the hash features, we find some channels exhibit meaningless noise. Via Fast Fourier Transform (FFT), we distinguish and mask thes… view at source ↗

Figures from the paper (3 more)

**Figure 5.** Figure 5: Visualization of Reconstruction Results. The SSIM is displayed in the [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]

**Figure 6.** Figure 6: The segment results of different methods on 3 datasets. [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

**Figure 8.** Figure 8: Ablation study of LN and MCI on NAF. 4 Conclusion In conclusion, we address the local-global training mismatch in NeRF-based CBCT reconstruction, where the hash encoder’s local sparse updates conflict with the neural network’s global dense updates, causing misaligned f…

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

HASD: Hierarchical Adaption for pathology Slide-level Domain-shift
cs.AI 2025-06 conditional novelty 6.0 of 10

A hierarchical domain adaptation framework improves slide-level HER2 grading and survival prediction across medical centers by aligning features at domain, slide, and patch levels.

Reference graph

Works this paper leans on

27 extracted references · 10 canonical work pages · cited by 1 Pith paper

[1]

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
[2]

Nature communications13(1), 4128 (2022)

Antonelli, M., Reinke, A., Bakas, S., Farahani, K., Kopp-Schneider, A., Landman, B.A., Litjens, G., Menze, B., Ronneberger, O., Summers, R.M., et al.: The medical segmentation decathlon. Nature communications13(1), 4128 (2022)

2022
[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]

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]

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]

(No Title) (2014)

Diederik, K.: Adam: A method for stochastic optimization. (No Title) (2014)

2014
[7]

Josa a 1(6), 612–619 (1984)

Feldkamp, L.A., Davis, L.C., Kress, J.W.: Practical cone-beam algorithm. Josa a 1(6), 612–619 (1984)

1984
[8]

In: The Thirty- eighth Annual Conference on Neural Information Processing Systems (2024)

Gao, Z., Planche, B., Zheng, M., Chen, X., Chen, T., Wu, Z.: Ddgs-ct: Direction- disentangled gaussian splatting for realistic volume rendering. In: The Thirty- eighth Annual Conference on Neural Information Processing Systems (2024)

work page 2024

Show all 27 references

[9]

arXiv preprint arXiv:2409.11169 (2024)

Guo, P., Zhao, C., Yang, D., Xu, Z., Nath, V., Tang, Y., Simon, B., Belue, M., Harmon, S., Turkbey, B., et al.: Maisi: Medical ai for synthetic imaging. arXiv preprint arXiv:2409.11169 (2024)

2024
[10]

Nature methods18(2), 203–211 (2021)

Isensee, F., Jaeger, P.F., Kohl, S.A., Petersen, J., Maier-Hein, K.H.: nnu-net: a self-configuring method for deep learning-based biomedical image segmentation. Nature methods18(2), 203–211 (2021)

2021
[11]

IEEE Access9, 71091–71103 (2021)

Jiao, F., Gui, Z., Li, K., Shangguang, H., Wang, Y., Liu, Y., Zhang, P.: A dual- domain cnn-based network for ct reconstruction. IEEE Access9, 71091–71103 (2021)

2021
[12]

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
[13]

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

Liu, J., Bai, X.: Volumenerf: Ct volume reconstruction from a single projection view. In: International Conference on Medical Image Computing and Computer- Assisted Intervention. pp. 743–753. Springer (2024)

2024
[14]

In: Eu- ropean conference on computer vision

Mildenhall, B., Srinivasan, P.P., Tancik, M., Barron, J.T., Ramamoorthi, R., Ng, R.: Nerf: Representing scenes as neural radiance fields for view synthesis. In: Eu- ropean conference on computer vision. pp. 405–421. Springer (2020)

2020
[15]

arXiv:2201.05989 (2022) 10 Xu et al

Müller, T., Evans, A., Schied, C., Keller, A.: Instant neural graphics primitives with a multiresolution hash encoding. arXiv:2201.05989 (2022) 10 Xu et al

2022 arXiv
[16]

NVIDIA, Vingelmann, P., Fitzek, F.H.: Cuda, release: 10.2.89 (2020), https:// developer.nvidia.com/cuda-toolkit

2020
[17]

In: Wallach, H., Larochelle, H., Beygelzimer, A., d'Alché-Buc, F., Fox, E., Garnett, R

Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., Antiga, L., Desmaison, A., Kopf, A., Yang, E., De- Vito, Z., Raison, M., Tejani, A., Chilamkurthy, S., Steiner, B., Fang, L., Bai, J., Chintala, S.: Pytorch: An imperat...

2019
[18]

Medical physics 50(3), 1917–1927 (2023)

Podobnik, G., Strojan, P., Peterlin, P., Ibragimov, B., Vrtovec, T.: Han-seg: The head and neck organ-at-risk ct and mr segmentation dataset. Medical physics 50(3), 1917–1927 (2023)

2023
[19]

The Cancer Imaging Archive (2016)

Roth, H., Farag, A., Turkbey, E., Lu, L., Liu, J., Summers, R.: Data from pancreas- ct (version 2)[data set]. The Cancer Imaging Archive (2016)

2016
[20]

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
[21]

Journal-Canadian Dental Association 72(1), 75 (2006)

Scarfe, W.C., Farman, A.G., Sukovic, P., et al.: Clinical applications of cone-beam computed tomography in dental practice. Journal-Canadian Dental Association 72(1), 75 (2006)

2006
[22]

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
[23]

Radiology: Artificial Intelligence 5(5), e230024 (2023)

Wasserthal, J., Breit, H.C., Meyer, M.T., Pradella, M., Hinck, D., Sauter, A.W., Heye, T., Boll, D.T., Cyriac, J., Yang, S., et al.: Totalsegmentator: robust segmen- tation of 104 anatomic structures in ct images. Radiology: Artificial Intelligence 5(5), e230024 (2023)

2023
[24]

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)

2021
[25]

arXiv preprint arXiv:2405.20693 (2024)

Zha, R., Lin, T.J., Cai, Y., Cao, J., Zhang, Y., Li, H.: R2-gaussian: Rectify- ing radiative gaussian splatting for tomographic reconstruction. arXiv preprint arXiv:2405.20693 (2024)

2024 arXiv
[26]

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

Zha, R., Zhang, Y., Li, H.: Naf: neural attenuation fields for sparse-view cbct reconstruction. In: International Conference on Medical Image Computing and Computer-Assisted Intervention. pp. 442–452. Springer (2022)

2022
[27]

ACS Applied Materials & Interfaces15(29), 35024–35033 (2023)

Zheng, Y., Hatzell, K.B.: Ultrasparse view x-ray computed tomography for 4d imaging. ACS Applied Materials & Interfaces15(29), 35024–35033 (2023)

2023

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [1]

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

[2] [2]

Nature communications13(1), 4128 (2022)

Antonelli, M., Reinke, A., Bakas, S., Farahani, K., Kopp-Schneider, A., Landman, B.A., Litjens, G., Menze, B., Ronneberger, O., Summers, R.M., et al.: The medical segmentation decathlon. Nature communications13(1), 4128 (2022)

2022

[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]

(No Title) (2014)

Diederik, K.: Adam: A method for stochastic optimization. (No Title) (2014)

2014

[7] [7]

Josa a 1(6), 612–619 (1984)

Feldkamp, L.A., Davis, L.C., Kress, J.W.: Practical cone-beam algorithm. Josa a 1(6), 612–619 (1984)

1984

[8] [8]

In: The Thirty- eighth Annual Conference on Neural Information Processing Systems (2024)

Gao, Z., Planche, B., Zheng, M., Chen, X., Chen, T., Wu, Z.: Ddgs-ct: Direction- disentangled gaussian splatting for realistic volume rendering. In: The Thirty- eighth Annual Conference on Neural Information Processing Systems (2024)

work page 2024