Pith. sign in

REVIEW 3 major objections 4 minor 72 references

Patient-Agnostic Synthetic Pretraining for Efficient Patient-Specific Intraoperative 2D/3D Registration

T0 review · 3 major / 4 minor · reviewed 2026-07-31 · deepseek-v4-flash

Pith's one-line read Patient-agnostic synthetic pretraining can cut the per-patient training cost of intraoperative 2D/3D registration by more than 40 times while preserving sub-millimeter accuracy, this paper argues.

desk verdict Solid, useful pretraining/adaptation study for patient-specific 2D/3D registration; the headline 44x speed-up only counts fine-tuning, not the one-time pretraining, which is never timed. read the letter →

arxiv 2607.23343 v1 pith:GYXYRRQQ submitted 2026-07-25 eess.IV cs.AIcs.CV

classification eess.IVcs.AIcs.CV
keywords 2D/3Dregistrationpatient-specificadaptationsyntheticpretrainingdomainrandomizationsphericalsimilaritylearningdifferentiableposerefinementintraoperativeimagingdigitallyreconstructedradiographs
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 tries to show that the expensive step in patient-specific 2D/3D registration—training a separate model from scratch for each new patient—can be replaced by a two-stage recipe: pretrain once on synthetic X-ray-like images (digitally reconstructed radiographs, DRRs) generated from many CT volumes, then fine-tune briefly on a small set of synthetic projections from the target patient's own CT. The central question is whether this shortcut preserves the accuracy of full per-patient training. The authors report that with full fine-tuning the pretrained model reaches 84.7% sub-millimeter success on pelvic fluoroscopy and 85.0% on cerebral angiography, statistically close to the 86.1% and 85.0% achieved from scratch, while cutting per-patient training time from about 20.7–20.9 hours to 0.47–0.48 hours—roughly a 44-fold reduction. A segmentation-free domain randomization scheme (intensity, projection physics, occlusion, field-of-view perturbations) is what lets the pretraining generalize to real X-rays without anatomical labels. If the claim holds, per-patient registration becomes an adaptation problem instead of a training problem, which matters for clinical deployment where speed matters.

What carries the argument

The framework combines three pieces: (1) patient-agnostic synthetic pretraining, in which DRRs generated from multiple CT volumes under randomized appearance, physics, occlusion, and field-of-view perturbations teach the network pose-sensitive features; (2) spherical similarity learning, in which image features are mapped to a hypersphere and the geodesic discrepancy between fixed and moving projections is minimized, with a bi-invariant SO(4) pose-gradient supervision aligning the learned metric's gradient to pose-space geodesic directions; and (3) differentiable Levenberg-Marquardt refinement, which iteratively updates the pose using Jacobians of the learned residual. The patient-specific a

What would settle it

Compute the total wall-clock time of pretraining plus patient-specific fine-tuning on the same GPU, then plot total time per patient against the number of patients treated. If the pretraining stage takes tens of hours, the break-even point may exceed realistic caseloads, which would overturn the paper's central efficiency claim. The from-scratch baseline time is given (20.7–20.9 h), so the comparison is directly testable.

Watch

Extended reading notes

Core claim

The paper's central claim is that a model pretrained on synthetic projections from multiple patients learns a pose-sensitive representation that transfers to a new patient, so that only a short fine-tuning step on the target CT's synthetic projections is needed. Direct transfer without adaptation fails (about 20% sub-millimeter success), but fine-tuning recovers nearly all of the from-scratch accuracy: on pelvic data, 10% of the usual fine-tuning budget reaches 42.4% SMSR, 50% reaches 74.3%, full fine-tuning reaches 84.7% versus 86.1% from scratch; on cerebral data, full fine-tuning matches 85.0% exactly. The authors attribute the transfer to the spherical similarity landscape and the SO(4)-

Load-bearing premise

The efficiency claim rests on the assumption that the one-time patient-agnostic pretraining cost is small enough to be amortized across the patients a site actually treats—the paper reports only fine-tuning time (0.47–0.48 h) and never the pretraining wall-clock time, so if pretraining is expensive and caseload is small, the full pipeline could cost more than training from scratch.

Editorial extensions

If this is right

  • Per-patient training time for learning-based 2D/3D registration drops from about 20.7–20.9 hours to under half an hour, making same-day adaptation between cases realistic.
  • A single patient-agnostic pretrained model can serve as a reusable starting point across many patients, so the one-time pretraining cost is amortized over the caseload.
  • Segmentation-free domain randomization provides robustness to real fluoroscopic appearance without requiring anatomical labels or segmentation masks, removing a major annotation bottleneck.
  • Fine-tuning only the pose regressor and similarity network recovers most of the accuracy of full fine-tuning, so the most efficient adaptation setting is partial fine-tuning.
  • The learned spherical metric supplies a differentiable landscape that supports Levenberg-Marquardt refinement, so final accuracy depends jointly on the quality of the initialization and the learned similarity.

Reading between the lines

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

  • Editorial extension: the reported 44x speed-up counts only patient-specific fine-tuning; if the one-time pretraining cost is large, the total time savings depend on the number of patients treated. A deployment site seeing few patients might not break even.
  • Editorial extension: the same pretraining-plus-adaptation logic could apply to other per-patient tasks, such as instrument tracking or reconstruction from sparse views, wherever a shared geometry prior can be learned from synthetic projections.
  • Editorial extension: the domain randomization pools could be tuned per anatomy; adding surgical-tool occlusion specifically for interventional workflows might further close the synthetic-to-real gap, though the paper does not test this.
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.

Referee Report

3 major / 4 minor

Summary. This manuscript proposes a two-stage strategy for patient-specific 2D/3D registration: pretrain on synthetic DRRs from multiple CT volumes with a segmentation-free domain randomization, then fine-tune on a small set of synthetic projections from the target CT. The architecture follows the authors' prior spherical similarity learning framework with CNN-Transformer encoders, an E-CNN volume module, and differentiable Levenberg–Marquardt refinement. On DeepFluoro and Ljubljana, full fine-tuning achieves 84.7%/85.0% SMSR versus 86.1%/85.0% trained from scratch, while reported patient-specific fine-tuning time drops from ~20.7/20.9 h to ~0.47/0.48 h (~44×). The paper also ablates fine-tuned modules and domain randomization pools.

Significance. The contribution is potentially valuable: it offers an annotation-free way to convert expensive per-patient training into lightweight adaptation, and the experiments use public datasets with a leave-one-out protocol. Strengths include the controlled comparison of architectures, the domain-randomization ablation, and the direct wall-clock measurement of fine-tuning time. However, the headline efficiency claim is only partially supported because the one-time pretraining cost is not reported, and all accuracy comparisons are point estimates. The central idea is reasonable and empirically defensible pending these measurements.

major comments (3)
  1. [§4.2, Table 2] The 44× speed-up compares only fine-tuning time (0.47/0.48 h) to from-scratch training (20.70/20.90 h). The pretraining stage, described in §4.2 as 500k pose-regressor + 200k similarity-network samples, is never given a wall-clock cost. The efficiency claim in the title and conclusion is therefore load-bearing on amortization: if pretraining costs about as much as one from-scratch run, the first patient is not faster. Please report pretraining wall-clock time and provide a break-even analysis (number of patients for which pretraining pays off), or qualify the claim as reducing per-patient adaptation time only.
  2. [Table 2 (and Tables 3–4)] All SMSR and mTRE numbers are point estimates from a single run. The claimed 'comparable accuracy' rests on 84.7% vs 86.1% on DeepFluoro and equality (85.0% vs 85.0%) on Ljubljana; without repeated-seed means, standard deviations, or a paired comparison, this difference cannot be assessed. Please report mean ± std over at least three seeds, or equivalent uncertainty quantification, for the main comparisons.
  3. [§3.3, Eq. (9)] The core spherical similarity learning and the bi-invariant SO(4) pose-gradient supervision are described by reference to the authors' conference paper [28]. Since these components are central to the method and the manuscript is intended as a journal version, the paper should be self-contained: define the spherical exponential map in Eq. (7), the SO(4) embedding, and the geodesic discrepancy used in Eq. (9), or include them in a supplement. As written, a reader cannot reproduce the training objective without obtaining [28].
minor comments (4)
  1. [Table 2] Define what '10% FT', '25% FT', etc. mean (e.g., fractions of the 20k-sample adaptation set). The 'Fine-tuning amount' column is ambiguous.
  2. [§4.2, Fig. 7] 'The learned deep similarity is visualized in Fig. 7' is a dangling reference; the figure should be described and interpreted in the text. Also add the missing period after 'Fig. 7'.
  3. [Eq. (18)] Define I0 and the valid intensity range after normalization; this affects reproducibility of the preprocessing.
  4. [§4.3] 'The baselines follow the same setting as we introduced in [28]' should be expanded so the reader does not need to consult [28] to understand the baseline evaluation protocol.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation: the central accuracy and speed-up claims are direct empirical measurements on external benchmarks; self-citations supply architecture, not the result.

full rationale

The paper's central claims are empirical rather than derived: patient-agnostic pretraining plus patient-specific fine-tuning is compared against from-scratch training on two public datasets (DeepFluoro, Ljubljana) under a leave-one-out protocol, with accuracy measured by SMSR and mTRE against ground-truth poses. There is no equation whose output is also its input, and no fitted parameter is relabeled as a prediction. The method does build on the authors' own conference work [28] ('The overall framework follows our conference version [28]'; 'Building on our previous spherical similarity learning framework [28]'), but this self-citation supplies architecture and training losses, not the measured conclusion, and the comparison is controlled by keeping the architecture identical between from-scratch and pretrained-fine-tuned settings. The main caveat is that the 44x speed-up compares only fine-tuning time (0.47 h) to from-scratch training time (20.70 h) and does not report the one-time patient-agnostic pretraining wall-clock cost, so the practical total-cost efficiency claim is incompletely supported. That is a missing measurement, not circularity: the reported claim is explicitly about 'patient-specific training time' and the metric is defined accordingly. Therefore no circular step is present; score is low only to acknowledge the unreported pretraining cost and reliance on prior self-cited components.

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

No new physical entities are introduced. The free parameters are hand-set training/optimization choices; the axioms are domain assumptions about DRR realism, representativeness of the small cohorts, and correctness of the inherited spherical similarity machinery.

free parameters (5)
  • Dataset-specific pose sampling ranges = DeepFluoro: rotations [-45,45] deg, translations x,z [-150,150] mm, y [-1000,-450] mm; Ljubljana: rotations alpha [-45,9
    Hand-set distributions for synthetic pose labels, taken from [40]; they define the capture range and difficulty of the training task.
  • Augmentation activation probabilities pI=pP=pM = 0.5 each
    Eq. (13) decides which domain-randomization pools are active per sample; chosen without a derivation or sensitivity study.
  • LM optimization schedule = 150 iterations, lr 5e-3, step decay 0.9 per 25 iterations, termination std < 1e-2
    Hand-set refinement hyperparameters that affect final mTRE/SMSR values.
  • Adaptation sample budget K = 20,000 (full); 10%, 25%, 50% in Table 2
    The default patient-specific fine-tuning budget; the headline 44x speed-up is computed at this budget, and the trade-off curve depends on it.
  • Pretraining data scale = 500k pose-regressor samples; 200k similarity-learning samples
    Chosen 'for consistency in training-data scale' rather than derived from an optimality argument; affects transferable representation quality.
assumptions (5)
  • domain assumption DRRs generated with Siddon ray-casting plus Eq. (18) logarithmic intensity inversion approximate clinical X-ray attenuation
    All synthetic pretraining and adaptation data are generated under this model; if the approximation fails, learned similarity will not transfer to real fluoroscopy.
  • domain assumption The leave-one-out CT pool is representative enough for transfer (6 pelvic CTs on DeepFluoro, 10 CBCTs on Ljubljana)
    Patient-agnostic pretraining utility depends on anatomical diversity across those cohorts; small datasets limit the generalization claim.
  • domain assumption Domain randomization pools G_I, G_P, G_M preserve pose-sensitive anatomy while removing appearance variability
    The method's robustness claim rests on this invariance-preservation property, asserted in §3.4 but not proven.
  • standard math Spherical similarity learning and bi-invariant SO(4) pose-gradient formulation from [28] are correct and transferable
    The refinement stage inherits these constructions from the authors' prior conference version without re-derivation here.
  • standard math Levenberg-Marquardt update Eq. (16) converges to a sufficiently good local optimum within 150 iterations
    Standard optimization assumption; the termination criterion and iteration budget are chosen by hand.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Patient-Agnostic Synthetic Pretraining for Efficient Patient-Specific Intraoperative 2D/3D Registration." pith.science (2026). https://pith.science/paper/GYXYRRQQ

@misc{pith2026260723343,
  author       = {Pith},
  title        = {Pith review of: Patient-Agnostic Synthetic Pretraining for Efficient Patient-Specific Intraoperative 2D/3D Registration},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GYXYRRQQ}},
  note         = {Machine review of arXiv:2607.23343}
}
read the original abstract

Intraoperative 2D/3D registration aligns preoperative CT volumes with intraoperative X-ray or fluoroscopic images and is essential for image-guided interventions. Recent learning-based and differentiable registration methods have shown promising accuracy, especially in patient-specific settings where abundant digitally reconstructed radiographs (DRRs) can be synthesized from the target CT. However, training a separate patient-specific model from scratch for every new patient is computationally inefficient and limits practical deployment. In this work, we propose an efficient patient-specific 2D/3D registration framework based on patient-agnostic synthetic pretraining and spherical similarity learning. The model is first pretrained on synthetic DRRs generated from multiple CT volumes to learn transferable pose-sensitive representations, and is then adapted to a new patient using only a limited number of synthetic projections from the target CT. To improve synthetic-to-real robustness without requiring anatomical labels, we introduce a segmentation-free domain randomization strategy that perturbs image intensity, projection physics, field-of-view, occlusion, and fluoroscopic artifacts. The adapted model provides an initial pose estimate, which is further refined using spherical similarity learning and differentiable Levenberg-Marquardt optimization. Experiments on multiple anatomical datasets evaluate whether patient-agnostic synthetic pretraining can improve the efficiency of patient-specific registration, with particular focus on the trade-off between adaptation cost and registration accuracy. The results demonstrate that patient-agnostic synthetic pretraining can significantly reduce patient-specific training requirements while preserving accurate intraoperative 2D/3D registration.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

72 extracted references · 2 linked inside Pith

  1. [40]

    https://arxiv.org/abs/2503.16309

    Gopalakrishnan, V., Dey, N., Chlorogian- nis, D.-D., Abumoussa, A., Larson, A.M., Orbach, D.B., Frisken, S., Golland, P.: Rapid patient-specific neural networks for intraop- erative X-ray to volume registration (2025). https://arxiv.org/abs/2503.16309

  2. [28]

    In: Proceedings of the IEEE/CVF Winter Conference on Applications of Com- puter Vision, pp

    Chen, M., Kong, Y.: Intraoperative 2d/3d registration via spherical similarity learning and differentiable levenberg-marquardt opti- mization. In: Proceedings of the IEEE/CVF Winter Conference on Applications of Com- puter Vision, pp. 7239–7249 (2026)

  3. [1]

    In: International Work- shop on Computer-Assisted and Robotic Endoscopy, pp

    Grupp, R.B., Armand, M., Taylor, R.H.: Patch-based image similarity for intraopera- tive 2d/3d pelvis registration during periac- etabular osteotomy. In: International Work- shop on Computer-Assisted and Robotic Endoscopy, pp. 153–163 (2018). Springer

  4. [2]

    IEEE transactions on medical imaging17(4), 586–595 (1998)

    Penney, G.P., Weese, J., Little, J.A., Desmedt, P., Hill, D.L.,et al.: A comparison of similarity measures for use in 2-d-3-d med- ical image registration. IEEE transactions on medical imaging17(4), 586–595 (1998)

  5. [3]

    In: International Con- ference on Medical Image Computing and Computer-Assisted Intervention, pp

    Knaan, D., Joskowicz, L.: Effective intensity- based 2d/3d rigid registration between fluo- roscopic x-ray and ct. In: International Con- ference on Medical Image Computing and Computer-Assisted Intervention, pp. 351–358 (2003). Springer

  6. [4]

    Medical Image Analysis67, 101815 (2021)

    Frysch, R., Pfeiffer, T., Rose, G.: A novel approach to 2d/3d registration of x-ray images using grangeat’s relation. Medical Image Analysis67, 101815 (2021)

  7. [5]

    Physics in Medicine & Biology61(8), 3009 (2016)

    De Silva, T., Uneri, A., Ketcha, M., Reaung- amornrat, S., Kleinszig, G., Vogt, S., Aygun, N., Lo, S., Wolinsky, J., Siewerdsen, J.: 3d– 2d image registration for target localization in spine surgery: investigation of similarity metrics providing robustness to content mis- match. Physics in Medicine & Biology61(8), 3009 (2016)

  8. [6]

    Medical image analysis16(3), 642–661 (2012)

    Markelj, P., Tomaˇ zeviˇ c, D., Likar, B., Pernuˇ s, F.: A review of 3d/2d registration meth- ods for image-guided interventions. Medical image analysis16(3), 642–661 (2012)

Show all 72 references
  1. [7]

    In: 2025 IEEE 22nd International Symposium on Biomedical Imaging (ISBI), pp

    Chen, M., Zhang, Z.: Introducing learning rate adaptation cma-es into rigid intraopera- tive 2d/3d registration for spinal surgery. In: 2025 IEEE 22nd International Symposium on Biomedical Imaging (ISBI), pp. 1–4 (2025). IEEE

  2. [8]

    arXiv preprint arXiv:2402.05642 (2024)

    Chen, M., Li, T., Zhang, Z., Kong, Y.: An optimization-based baseline for rigid 2d/3d registration applied to spine surgi- cal navigation using cma-es. arXiv preprint arXiv:2402.05642 (2024)

  3. [9]

    IEEE transactions on medical imaging37(1), 262–272 (2017)

    Toews, M., Wells, W.M.: Phantomless auto- calibration and online calibration assessment for a tracked freehand 2-d ultrasound probe. IEEE transactions on medical imaging37(1), 262–272 (2017)

  4. [10]

    Cambridge NA Report NA2009/06, University of Cambridge, Cam- bridge26, 26–46 (2009)

    Powell, M.J.,et al.: The bobyqa algo- rithm for bound constrained optimization without derivatives. Cambridge NA Report NA2009/06, University of Cambridge, Cam- bridge26, 26–46 (2009)

  5. [11]

    Evolutionary computation9(2), 159–195 (2001)

    Hansen, N., Ostermeier, A.: Completely derandomized self-adaptation in evolution strategies. Evolutionary computation9(2), 159–195 (2001)

  6. [12]

    Medical Physics52(7), 17896 (2025)

    Li, T., Chen, M., Li, M., Li, C., Kong, Y.: Automatic x-ray to ct registration using 15 embedding reconstruction and lite cross- attention. Medical Physics52(7), 17896 (2025)

  7. [13]

    IEEE transactions on medical imaging 35(5), 1352–1363 (2016)

    Miao, S., Wang, Z.J., Liao, R.: A cnn regres- sion approach for real-time 2d/3d registra- tion. IEEE transactions on medical imaging 35(5), 1352–1363 (2016)

  8. [14]

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

    Leroy, A., Cafaro, A., Gessain, G., Cham- pagnac, A., Gr´ egoire, V., Deutsch, E., Lepetit, V., Paragios, N.: Structuregnet: Structure-guided multimodal 2d-3d registra- tion. In: International Conference on Medical Image Computing and Computer-Assisted Intervention, pp. 771–78...

  9. [15]

    Biomedical Signal Processing and Control95, 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 Control95, 106384 (2024)

  10. [16]

    In: Interna- tional Conference on Medical Image Comput- ing and Computer Assisted Intervention, pp

    Esteban, J., Grimm, M., Unberath, M., Zahnd, G., Navab, N.: Towards fully auto- matic x-ray to ct registration. In: Interna- tional Conference on Medical Image Comput- ing and Computer Assisted Intervention, pp. 631–639 (2019). Springer

  11. [17]

    IEEE transactions on medical imaging40(9), 2221–2232 (2021)

    Grimm, M., Esteban, J., Unberath, M., Navab, N.: Pose-dependent weights and domain randomization for fully automatic x- ray to ct registration. IEEE transactions on medical imaging40(9), 2221–2232 (2021)

  12. [18]

    In: International Workshop on Biomed- ical Image Registration, pp

    Brandst¨ atter, S., Seeb¨ ock, P., F¨ urb¨ ock, C., Pochepnia, S., Prosch, H., Langs, G.: Rigid single-slice-in-volume registration via rotation-equivariant 2d/3d feature match- ing. In: International Workshop on Biomed- ical Image Registration, pp. 280–294 (2024). Springer

  13. [19]

    In: International Conference on Medi- cal Image Computing and Computer-assisted Intervention, pp

    Markova, V., Ronchetti, M., Wein, W., Zettinig, O., Prevost, R.: Global multi-modal 2d/3d registration via local descriptors learn- ing. In: International Conference on Medi- cal Image Computing and Computer-assisted Intervention, pp. 269–279 (2022). Springer

  14. [20]

    In: Interna- tional Conference on Medical Image Comput- ing and Computer Assisted Intervention, pp

    Unberath, M., Zaech, J.-N., Lee, S.C., Bier, B., Fotouhi, J., Armand, M., Navab, N.: Deepdrr–a catalyst for machine learning in fluoroscopy-guided procedures. In: Interna- tional Conference on Medical Image Comput- ing and Computer Assisted Intervention, pp. 98–106 (2018). Springer

  15. [21]

    In: Proceedings of the IEEE/CVF Winter Con- ference on Applications of Computer Vision, pp

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

  16. [22]

    In: Proceed- ings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition, pp

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

  17. [23]

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

    Gao, C., Liu, X., Gu, W., Killeen, B., Armand, M., Taylor, R., Unberath, M.: Gen- eralizing spatial transformers to projective geometry with applications to 2d/3d registra- tion. In: International Conference on Medical Image Computing and Computer Assisted Intervention, pp. 32...

  18. [24]

    IEEE transactions on medical imaging (2023)

    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)

  19. [25]

    In: 2024 IEEE International Symposium on Biomedical Imaging (ISBI), pp

    Chen, M., Zhang, Z., Gu, S., Ge, Z., Kong, Y.: Fully differentiable correlation-driven 2d/3d registration for x-ray to ct image fusion. In: 2024 IEEE International Symposium on Biomedical Imaging (ISBI), pp. 1–5 (2024)

  20. [26]

    In: ICASSP 2024-2024 IEEE International Con- ference on Acoustics, Speech and Signal Processing (ICASSP), pp

    Chen, M., Zhang, Z., Gu, S., Kong, Y.: Embedded feature similarity optimiza- tion with specific parameter initialization for 2d/3d medical image registration. In: ICASSP 2024-2024 IEEE International Con- ference on Acoustics, Speech and Signal Processing (ICASSP), pp. 1521–152...

  21. [27]

    In: Interna- tional Conference on Medical Image Comput- ing and Computer-Assisted Intervention, pp

    Zhang, B., Faghihroohi, S., Azampour, M.F., 16 Liu, S., Ghotbi, R., Schunkert, H., Navab, N.: A patient-specific self-supervised model for automatic x-ray/ct registration. In: Interna- tional Conference on Medical Image Comput- ing and Computer-Assisted Intervention, pp. 515–5...

  22. [29]

    In: International Con- ference on Medical Image Computing and Computer Assisted Intervention, pp

    Guo, H., Xu, X., Xu, S., Wood, B.J., Yan, P.: End-to-end ultrasound frame to volume registration. In: International Con- ference on Medical Image Computing and Computer Assisted Intervention, pp. 56–65 (2021). Springer

  23. [30]

    In: 2016 IEEE 13th International Sym- posium on Biomedical Imaging (ISBI), pp

    Miao, S., Wang, Z.J., Zheng, Y., Liao, R.: Real-time 2d/3d registration via cnn regres- sion. In: 2016 IEEE 13th International Sym- posium on Biomedical Imaging (ISBI), pp. 1430–1434 (2016). IEEE

  24. [31]

    In: In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Liao, H., Lin, W.-A., Zhang, J., Zhang, J., Luo, J., Zhou, S.K.: Multiview 2d/3d rigid registration via a point-of-interest network for tracking and triangulation. In: In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 12638–12647 (2019)

  25. [32]

    In: Inter- national Conference on Medical Image Com- puting and Computer-Assisted Intervention, pp

    Shrestha, P., Xie, C., Shishido, H., Yoshii, Y., Kitahara, I.: X-ray to ct rigid registration using scene coordinate regression. In: Inter- national Conference on Medical Image Com- puting and Computer-Assisted Intervention, pp. 781–790 (2023). Springer

  26. [33]

    In: Proceedings of the Asian Conference on Computer Vision, pp

    Shrestha, P., Xie, C., Yoshii, Y., Kitahara, I.: Rayemb: Arbitrary landmark detection in x-ray images using ray embedding subspace. In: Proceedings of the Asian Conference on Computer Vision, pp. 665–681 (2024)

  27. [34]

    IEEE transactions on pattern analysis and machine intelligence34(7), 1444–1450 (2012)

    Li, S., Xu, C., Xie, M.: A robust o (n) solution to the perspective-n-point problem. IEEE transactions on pattern analysis and machine intelligence34(7), 1444–1450 (2012)

  28. [35]

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

    Jaganathan, S., Wang, J., Borsdorf, A., Shetty, K., Maier, A.: Deep iterative 2d/3d registration. In: International Conference on Medical Image Computing and Computer- Assisted Intervention, pp. 383–392 (2021). Springer

  29. [36]

    IEEE transactions on medical imaging39(10), 3159–3174 (2020)

    Schaffert, R., Wang, J., Fischer, P., Borsdorf, A., Maier, A.: Learning an attention model for robust 2-d/3-d registration using point-to- plane correspondences. IEEE transactions on medical imaging39(10), 3159–3174 (2020)

  30. [37]

    IEEE transactions on medical imaging 36(9), 1939–1954 (2017)

    Wang, J., Schaffert, R., Borsdorf, A., Heigl, B., Huang, X., Hornegger, J., Maier, A.: Dynamic 2-d/3-d rigid registration frame- work using point-to-plane correspondence model. IEEE transactions on medical imaging 36(9), 1939–1954 (2017)

  31. [38]

    Communications of the ACM24(6), 381–395 (1981)

    Fischler, M.A., Bolles, R.C.: Random sam- ple consensus: a paradigm for model fitting with applications to image analysis and auto- mated cartography. Communications of the ACM24(6), 381–395 (1981)

  32. [39]

    International Journal of Com- puter Assisted Radiology and Surgery, 1–10 (2025)

    Downs, C., Sluijs, P., Cornelissen, S.A., Nijenhuis, F.t., Zwam, W.H.v., Gopalakrish- nan, V., Zhang, X., Su, R., Walsum, T.v.: Improving automatic cerebral 3d-2d cta-dsa registration. International Journal of Com- puter Assisted Radiology and Surgery, 1–10 (2025)

  33. [41]

    In: Workshop on Clinical Image-Based Procedures, pp

    Gopalakrishnan, V., Golland, P.: Fast auto- differentiable digitally reconstructed radio- graphs for solving inverse problems in intra- operative imaging. In: Workshop on Clinical Image-Based Procedures, pp. 1–11 (2022). Springer

  34. [42]

    In: 17 International Conference on Medical Imag- ing with Deep Learning, pp

    Toth, D., Cimen, S., Ceccaldi, P., Kurzendor- fer, T., Rhode, K., Mountney, P.: Training deep networks on domain randomized syn- thetic x-ray data for cardiac interventions. In: 17 International Conference on Medical Imag- ing with Deep Learning, pp. 468–482 (2019). PMLR

  35. [43]

    Imaging Neuroscience2, 2 (2024)

    Gopinath, K., Hoopes, A., Alexander, D.C., Arnold, S.E., Balbastre, Y., Billot, B., Casamitjana, A., Cheng, Y., Chua, R.Y.Z., Edlow, B.L.,et al.: Synthetic data in general- izable, learning-based neuroimaging. Imaging Neuroscience2, 2 (2024)

  36. [44]

    In: The Thirteenth International Conference on Learning Representations (2025)

    Dey, N., Billot, B., Wong, H.E., Wang, C., Ren, M., Grant, E., Dalca, A.V., Golland, P.: Learning general-purpose biomedical volume representations using randomized synthesis. In: The Thirteenth International Conference on Learning Representations (2025)

  37. [45]

    Medical Image Analysis105, 103669 (2025)

    Fu, J., Zheng, Y., Dey, N., Ferreira, D., Moreno, R.: Synthesizing individualized aging brains in health and disease with gener- ative models and parallel transport. Medical Image Analysis105, 103669 (2025)

  38. [46]

    Medical Image Analysis86, 102789 (2023)

    Billot, B., Greve, D.N., Van Leemput, K., Fischl, B., Iglesias, J.E., Dalca, A.V.: Synth- seg: Segmentation of brain mri scans of any contrast and resolution without retraining. Medical Image Analysis86, 102789 (2023)

  39. [47]

    Proceedings of the National Academy of Sci- ences120(9), 2216399120 (2023)

    Billot, B., Magdamo, C., Cheng, Y., Arnold, S.E., Das, S., Iglesias, J.E.: Robust machine learning segmentation for large-scale analysis of heterogeneous clinical brain mri datasets. Proceedings of the National Academy of Sci- ences120(9), 2216399120 (2023)

  40. [48]

    In: 2024 IEEE International Sympo- sium on Biomedical Imaging (ISBI), pp

    Laso, P., Cerri, S., Sorby-Adams, A., Guo, J., Mateen, F., Goebl, P., Wu, J., Liu, P., Li, H.B., Young, S.I.,et al.: Quantifying white matter hyperintensity and brain volumes in heterogeneous clinical and low-field portable mri. In: 2024 IEEE International Sympo- sium on Biome...

  41. [49]

    NeuroImage260, 119474 (2022)

    Hoopes, A., Mora, J.S., Dalca, A.V., Fischl, B., Hoffmann, M.: Synthstrip: skull-stripping for any brain image. NeuroImage260, 119474 (2022)

  42. [50]

    IEEE Trans- actions on Medical Imaging41(3), 543–558 (2022)

    Hoffmann, M., Billot, B., Greve, D.N., Igle- sias, J.E., Fischl, B., Dalca, A.V.: Synth- morph: learning contrast-invariant registra- tion without acquired images. IEEE Trans- actions on Medical Imaging41(3), 543–558 (2022)

  43. [51]

    Imaging Neuroscience2, 1–33 (2024)

    Hoffmann, M., Hoopes, A., Greve, D.N., Fis- chl, B., Dalca, A.V.: Anatomy-aware and acquisition-agnostic joint registration with synthmorph. Imaging Neuroscience2, 1–33 (2024)

  44. [52]

    International journal of computer assisted radiology and surgery14(9), 1517–1528 (2019)

    Unberath, M., Zaech, J.-N., Gao, C., Bier, B., Goldmann, F., Lee, S.C., Fotouhi, J., Tay- lor, R., Armand, M., Navab, N.: Enabling machine learning in x-ray-based procedures via realistic simulation of image formation. International journal of computer assisted radiology and s...

  45. [53]

    Nature Machine Intelli- gence5(3), 294–308 (2023)

    Gao, C., Killeen, B.D., Hu, Y., Grupp, R.B., Taylor, R.H., Armand, M., Unberath, M.: Synthetic data accelerates the development of generalizable learning-based algorithms for x-ray image analysis. Nature Machine Intelli- gence5(3), 294–308 (2023)

  46. [54]

    Interna- tional Journal of Computer Assisted Radiol- ogy and Surgery19(5), 939–950 (2024)

    Huang, Z., Li, H., Shao, S., Zhu, H., Hu, H., Cheng, Z., Wang, J., Kevin Zhou, S.: Pele scores: pelvic x-ray landmark detection with pelvis extraction and enhancement. Interna- tional Journal of Computer Assisted Radiol- ogy and Surgery19(5), 939–950 (2024)

  47. [55]

    In: European Con- ference on Computer Vision, pp

    Zakharov, S., Ambrus ,, R., Guizilini, V., Kehl, W., Gaidon, A.: Photo-realistic neural domain randomization. In: European Con- ference on Computer Vision, pp. 310–327 (2022). Springer

  48. [56]

    In: Interna- tional Conference on Information Process- ing in Medical Imaging, pp

    Qin, C., Shi, B., Liao, R., Mansi, T., Rueck- ert, D., Kamen, A.: Unsupervised deformable registration for multi-modal images via disentangled representations. In: Interna- tional Conference on Information Process- ing in Medical Imaging, pp. 249–261 (2019). Springer

  49. [57]

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

    Grzech, D., Azampour, M.F., Glocker, B., 18 Schnabel, J., Navab, N., Kainz, B., Le Fol- goc, L.: A variational bayesian method for similarity learning in non-rigid image regis- tration. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. ...

  50. [58]

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

    Ronchetti, M., Wein, W., Navab, N., Zettinig, O., Prevost, R.: Disa: Differentiable similar- ity approximation for universal multimodal registration. In: International Conference on Medical Image Computing and Computer- Assisted Intervention, pp. 761–770 (2023). Springer

  51. [59]

    In: International Workshop on Biomedical Image Registration, pp

    Grzech, D., Folgoc, L.L., Azampour, M.F., Vlontzos, A., Glocker, B., Navab, N., Schn- abel, J., Kainz, B.: Unsupervised similarity learning for image registration with energy- based models. In: International Workshop on Biomedical Image Registration, pp. 229–240 (2024). Springer

  52. [60]

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

    Sideri-Lampretsa, V., Zimmer, V.A., Qiu, H., Kaissis, G., Rueckert, D.: Mad: Modality agnostic distance measure for image registra- tion. In: International Conference on Medical Image Computing and Computer-Assisted Intervention, pp. 147–156 (2023). Springer

  53. [61]

    In: Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Mok, T.C., Li, Z., Bai, Y., Zhang, J., Liu, W., Zhou, Y.-J., Yan, K., Jin, D., Shi, Y., Yin, X.,et al.: Modality-agnostic structural image representation learning for deformable multi- modality medical image registration. In: Pro- ceedings of the IEEE/CVF Conference on Compute...

  54. [62]

    In: Machine Learn- ing in Medical Imaging: 11th International Workshop (2020)

    Gu, W., Gao, C., Grupp, R., Fotouhi, J., Unberath, M.: Extended capture range of rigid 2d/3d registration by estimating rie- mannian pose gradients. In: Machine Learn- ing in Medical Imaging: 11th International Workshop (2020). Springer

  55. [63]

    IEEE transactions on medical imaging (2024)

    Billot, B., Dey, N., Moyer, D., Hoffmann, M., Turk, E.A., Gagoski, B., Grant, P.E., Golland, P.: Se (3)-equivariant and noise- invariant 3d rigid motion tracking in brain mri. IEEE transactions on medical imaging (2024)

  56. [64]

    Moyer, D., Abaci Turk, E., Grant, P.E., Wells, W.M., Golland, P.: Equivariant fil- ters for efficient tracking in 3d imaging. In: Medical Image Computing and Com- puter Assisted Intervention–MICCAI 2021: 24th International Conference, Strasbourg, France, September 27–October 1...

  57. [65]

    In: 2017 IEEE Winter Conference on Applications of Com- puter Vision (W ACV), pp

    Bui, M., Albarqouni, S., Schrapp, M., Navab, N., Ilic, S.: X-ray posenet: 6 dof pose estima- tion for mobile x-ray devices. In: 2017 IEEE Winter Conference on Applications of Com- puter Vision (W ACV), pp. 1036–1044 (2017). IEEE

  58. [66]

    International jour- nal of computer assisted radiology and surgery15, 759–769 (2020)

    Grupp, R.B., Unberath, M., Gao, C., Hege- man, R.A., Murphy, R.J., Alexander, C.P., Otake, Y., McArthur, B.A., Armand, M., Taylor, R.H.: Automatic annotation of hip anatomy in fluoroscopy for robust and effi- cient 2d/3d registration. International jour- nal of computer assist...

  59. [67]

    IEEE Transactions on Medi- cal Imaging32(8), 1550–1563 (2013)

    Mitrovi´ c, U., S˘piclin, Z., Likar, B., Per- nus˘, F.: 3d-2d registration of cerebral angiograms: A method and evaluation on clinical images. IEEE Transactions on Medi- cal Imaging32(8), 1550–1563 (2013)

  60. [68]

    Machine Learning and the Physical Sciences, NeurIPS 2024 (2024)

    Momeni, M., Gopalakrishnan, V., Dey, N., Golland, P., Frisken, S.: Differentiable voxel- based x-ray rendering improves sparse-view 3d cbct reconstruction. Machine Learning and the Physical Sciences, NeurIPS 2024 (2024)

  61. [69]

    Medical physics12(2), 252–255 (1985)

    Siddon, R.L.: Fast calculation of the exact radiological path for a three-dimensional ct array. Medical physics12(2), 252–255 (1985)

  62. [70]

    IEEE Transactions on Biomedical Engineering (2026)

    Kong, Y., Li, H., Zhang, S., Chen, M.: Pro-tuning: Prototype tuning of foundation models for volumetric medical image seg- mentation. IEEE Transactions on Biomedical Engineering (2026)

  63. [71]

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

    Lin, X., Xiang, Y., Yu, L., Yan, Z.: Beyond 19 adapting sam: Towards end-to-end ultra- sound image segmentation via auto prompt- ing. In: International Conference on Medical Image Computing and Computer-Assisted Intervention, pp. 24–34 (2024). Springer

  64. [72]

    Medical Image Analysis98, 103310 (2024) 20

    Chen, C., Miao, J., Wu, D., Zhong, A., Yan, Z., Kim, S., Hu, J., Liu, Z., Sun, L., Li, X., et al.: Ma-sam: Modality-agnostic sam adap- tation for 3d medical image segmentation. Medical Image Analysis98, 103310 (2024) 20

Pith tools

Reviewed July 31, 2026 · model on record in the stance chip above.