pith. sign in

arxiv: 2511.14286 · v2 · pith:5G3ON2LWnew · submitted 2025-11-18 · 💻 cs.CV

NeuralBoneReg: An Instance-Specific Label-Free Point Cloud-Based Method for Multi-Modal Bone Surface Registration

Luohong Wu , Matthias Seibold , Nicola A. Cavalcanti , Yunke Ao , Roman Flepp , Aidana Massalimova , Lilian Calvet , Philipp F\"urnstahl This is my paper

Pith reviewed 2026-05-25 07:39 UTC · model grok-4.3

classification 💻 cs.CV
keywords bone registrationpoint cloudself-supervisedunsigned distance fieldmulti-modalorthopedic surgeryneural implicitrigid transformation
0
0 comments X

The pith

NeuralBoneReg performs accurate multi-modal bone surface registration in a self-supervised, instance-specific manner using point clouds and neural distance fields.

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

This paper introduces a framework to align bone surfaces extracted from preoperative and intraoperative scans of different types. It learns an implicit model of the bone from one scan and finds the transformation to match the other scan's points without any labeled examples or data from other patients. Such registration is essential for transferring surgical plans accurately during orthopedic procedures. The approach uses a neural unsigned distance field for the bone shape and a multilayer perceptron to optimize the alignment. Results on three datasets show it performs as well as or better than existing methods that require supervision.

Core claim

The central discovery is that an instance-specific neural unsigned distance field representing the preoperative bone surface, paired with an MLP registration module, can recover rigid transformations to align intraoperative point clouds from heterogeneous modalities, achieving the reported error levels on the evaluated datasets without inter-subject training data.

What carries the argument

Implicit neural unsigned distance field (UDF) for the preoperative bone combined with MLP-based hypothesis generation for rigid transformations.

If this is right

  • It operates without inter-subject training data, relying only on the specific instance.
  • It achieves mean RRE/RTE of 1.83°/2.02 mm on UltraBones100k dataset.
  • It generalizes to hip and spine anatomies with errors of 1.90°/1.56 mm and 3.78°/2.80 mm respectively.
  • It supports both global initialization and local refinement stages.

Where Pith is reading between the lines

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

  • Surgeons could potentially use this to reduce reliance on manual alignment steps.
  • Extension to non-rigid deformations might be possible if the UDF representation is adapted accordingly.
  • The method's performance on very sparse point clouds remains to be tested in practice.
  • Integration into existing CAOS software could streamline workflow if inference times are acceptable.

Load-bearing premise

Point clouds derived from different modalities provide sufficiently complete and consistent surface information for the neural model to determine the correct rigid alignment without correspondences.

What would settle it

A test on a new dataset where the intraoperative point cloud is significantly incomplete or noisy, resulting in registration errors exceeding 5 degrees or 5 mm, would falsify the claim of robust cross-modal performance.

Figures

Figures reproduced from arXiv: 2511.14286 by Aidana Massalimova, Lilian Calvet, Luohong Wu, Matthias Seibold, Nicola A. Cavalcanti, Philipp F\"urnstahl, Roman Flepp, Yunke Ao.

Figure 1
Figure 1. Figure 1: Network architecture of the two-staged approach of NeuralBoneReg. In the [PITH_FULL_IMAGE:figures/full_fig_p011_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Recall–rate curves for UltraBones100k (first column), UltraBones-Hip (second [PITH_FULL_IMAGE:figures/full_fig_p020_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Exemplary qualitative results on UltraBones100k. The preoperative point cloud [PITH_FULL_IMAGE:figures/full_fig_p021_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Exemplary qualitative results on UltraBones-Hip. The preoperative point cloud [PITH_FULL_IMAGE:figures/full_fig_p022_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Exemplary qualitative results on SpineDepth. The preoperative point cloud [PITH_FULL_IMAGE:figures/full_fig_p023_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Plots of mean RTE and RRE versus depth of the shared backbone within the [PITH_FULL_IMAGE:figures/full_fig_p024_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Plots of mean RTE and RRE versus head counts on the SpineDepth dataset. [PITH_FULL_IMAGE:figures/full_fig_p024_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Example UDF traces of all heads during the training process for three datasets: [PITH_FULL_IMAGE:figures/full_fig_p025_8.png] view at source ↗
read the original abstract

In computer- and robot-assisted orthopedic surgery (CAOS), patient-specific surgical plans derived from preoperative imaging define target locations and implant trajectories. During surgery, these plans must be accurately transferred, relying on precise cross-registration between preoperative and intraoperative data. However, substantial modality heterogeneity across imaging modalities makes this registration challenging and error-prone. Robust, automatic, and modality-agnostic bone surface registration is therefore clinically important. We propose NeuralBoneReg, a self-supervised, surface-based framework that registers bone surfaces using 3D point clouds as a modality-agnostic representation. NeuralBoneReg includes two modules: an implicit neural unsigned distance field (UDF) that learns the preoperative bone model, and an MLP-based registration module that performs global initialization and local refinement by generating transformation hypotheses to align the intraoperative point cloud with the neural UDF. Unlike SOTA supervised methods, NeuralBoneReg operates in a self-supervised manner, without requiring inter-subject training data. We evaluated NeuralBoneReg against baseline methods on two publicly available multi-modal datasets: a CT-ultrasound dataset of the fibula and tibia (UltraBones100k) and a CT-RGB-D dataset of spinal vertebrae (SpineDepth). The evaluation also includes a newly introduced CT-ultrasound dataset of cadaveric subjects containing femur and pelvis (UltraBones-Hip), which will be made publicly available. NeuralBoneReg matches or surpasses existing methods across all datasets, achieving mean RRE/RTE of 1.83{\deg}/2.02 mm on UltraBones100k, 1.90{\deg}/1.56 mm on UltraBones-Hip, and 3.78{\deg}/2.80 mm on SpineDepth. These results demonstrate strong generalizability across anatomies and modalities, providing robust and accurate cross-modal alignment for CAOS.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper proposes NeuralBoneReg, a self-supervised, instance-specific framework for multi-modal bone surface registration in CAOS. It represents preoperative bone surfaces via an implicit neural unsigned distance field (UDF) learned from point clouds and uses an MLP to generate transformation hypotheses for global initialization and local refinement, aligning intraoperative point clouds without inter-subject labels or explicit correspondences. The method is evaluated on UltraBones100k (CT-US), a new UltraBones-Hip dataset (CT-US of femur/pelvis), and SpineDepth (CT-RGB-D), claiming to match or surpass baselines with mean RRE/RTE of 1.83°/2.02 mm, 1.90°/1.56 mm, and 3.78°/2.80 mm respectively.

Significance. If the results hold, the work offers a clinically relevant advance by enabling modality-agnostic registration without large labeled inter-subject datasets, a strength given the heterogeneity in CAOS imaging. The self-supervised UDF+MLP design and public release of UltraBones-Hip are positive contributions that could reduce reliance on supervised training.

major comments (2)
  1. [Abstract] Abstract: the central performance claim (matching or surpassing SOTA with the listed RRE/RTE values) rests on the unverified precondition that point clouds extracted from ultrasound and RGB-D are sufficiently complete, dense, and geometrically consistent with CT surfaces for the neural UDF to yield a usable loss landscape and for the MLP to converge reliably; no analysis, ablation, or discussion of partial coverage, density variation, or modality artifacts is provided to support this.
  2. [Results] Results (as summarized): the reported mean errors lack standard deviations, error bars, statistical significance tests against baselines, or details on data exclusion rules, making it impossible to verify whether the competitive numbers are robust or driven by favorable cases.
minor comments (1)
  1. [Abstract] The abstract and method description would benefit from explicit notation for the UDF loss and MLP hypothesis generation to clarify the self-supervised objective.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive feedback. We address the two major comments point by point below, indicating where revisions will be made to strengthen the manuscript. Both comments identify areas where additional analysis or reporting would improve clarity and verifiability.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central performance claim (matching or surpassing SOTA with the listed RRE/RTE values) rests on the unverified precondition that point clouds extracted from ultrasound and RGB-D are sufficiently complete, dense, and geometrically consistent with CT surfaces for the neural UDF to yield a usable loss landscape and for the MLP to converge reliably; no analysis, ablation, or discussion of partial coverage, density variation, or modality artifacts is provided to support this.

    Authors: We agree that the abstract's performance claims implicitly rely on the quality of the extracted intraoperative point clouds. The full manuscript describes the point-cloud extraction pipelines for each dataset (UltraBones100k, UltraBones-Hip, SpineDepth) and notes that these are real clinical acquisitions, but it does not include dedicated analysis of coverage, density variation, or artifact effects. To address this, we will add a new subsection in the Experiments section that (i) quantifies point-cloud coverage and density statistics across the test cases, (ii) discusses modality-specific artifacts observed in the data, and (iii) reports an ablation on registration accuracy under controlled down-sampling and partial-view conditions. This will make the precondition explicit and provide supporting evidence. revision: yes

  2. Referee: [Results] Results (as summarized): the reported mean errors lack standard deviations, error bars, statistical significance tests against baselines, or details on data exclusion rules, making it impossible to verify whether the competitive numbers are robust or driven by favorable cases.

    Authors: The current manuscript reports only mean RRE/RTE values. We acknowledge that the absence of standard deviations, error bars, statistical tests, and explicit data-exclusion criteria limits assessment of robustness. In the revised version we will (i) report mean ± standard deviation for all metrics, (ii) include error-bar plots in the result figures, (iii) add paired statistical significance tests (e.g., Wilcoxon signed-rank) against the baselines, and (iv) state the data-exclusion rules (if any) applied during evaluation. These additions will be placed in the Results and supplementary material. revision: yes

Circularity Check

0 steps flagged

No circularity; self-supervised method and dataset evaluations remain independent

full rationale

The paper presents NeuralBoneReg as a self-supervised framework (neural UDF for preoperative model + MLP hypothesis generator for registration) and reports empirical registration errors on held-out public and newly introduced datasets. No equations reduce the reported RRE/RTE values to quantities fitted from the evaluation data itself, no self-citation chain bears the central claim, and the modality-agnostic point-cloud assumption is stated as a precondition rather than derived by construction from the results. The derivation chain is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The performance claims rest on standard assumptions about neural network capacity to represent bone surfaces and the sufficiency of rigid transformations for alignment; no new physical entities are introduced.

free parameters (1)
  • UDF and MLP network hyperparameters
    Architecture sizes, learning rates, and training schedule are selected to enable the self-supervised alignment on each instance.
axioms (2)
  • domain assumption Neural networks can learn a faithful unsigned distance field representation of bone geometry from a single preoperative scan
    Invoked by the choice of implicit UDF module for the preoperative model.
  • domain assumption Point clouds from different modalities share a common rigid transformation to the preoperative surface
    Underlying the registration module's hypothesis generation.

pith-pipeline@v0.9.0 · 5902 in / 1399 out tokens · 35409 ms · 2026-05-25T07:39:41.576430+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

90 extracted references · 90 canonical work pages · 1 internal anchor

  1. [1]

    Cleary, T

    K. Cleary, T. M. Peters, Image-guided interventions: technology review and clinical applications, Annual review of biomedical engineering 12 (1) (2010) 119–142. 28

    work page 2010

  2. [2]

    Evrard, T

    R. Evrard, T. Schubert, L. Paul, P.-L. Docquier, Quality of resection margin with patient specific instrument for bone tumor resection, Jour- nal of bone oncology 34 (2022) 100434

    work page 2022

  3. [3]

    Sugano, Computer-assisted orthopedic surgery, Journal of Or- thopaedic Science 8 (2003) 442–448

    N. Sugano, Computer-assisted orthopedic surgery, Journal of Or- thopaedic Science 8 (2003) 442–448

    work page 2003

  4. [4]

    Baraza, C

    N. Baraza, C. Chapman, S. Zakani, K. Mulpuri, 3d-printed patient spe- cific instrumentation in corrective osteotomy of the femur and pelvis: a review of the literature, 3D Printing in Medicine 6 (2020) 1–7

    work page 2020

  5. [5]

    Scheufler, J

    K.-M. Scheufler, J. Franke, A. Eckardt, H. Dohmen, Accuracy of image-guided pedicle screw placement using intraoperative computed tomography-based navigation with automated referencing, part i: cer- vicothoracic spine, Neurosurgery 69 (4) (2011) 782–795

    work page 2011

  6. [6]

    E. M. Kanlić, F. DeLaRosa, M. Pirela-Cruz, Computer assisted or- thopaedic surgery–caos, Bosnian journal of basic medical sciences 6 (1) (2006) 7

    work page 2006

  7. [7]

    Picard, A

    F. Picard, A. H. Deakin, P. E. Riches, K. Deep, J. Baines, Computer assisted orthopaedic surgery: past, present and future, Medical engi- neering & physics 72 (2019) 55–65

    work page 2019

  8. [8]

    Gautam, S

    D. Gautam, S. Vivekanandan, M. D. Mazur, Robotic spine surgery: Systematic review of common error types and best practices, Operative Neurosurgery 28 (3) (2025) 295–302

    work page 2025

  9. [9]

    Wandvik, M

    C. Wandvik, M. E. Greil, S. Colby, D. Gautam, M. D. Mazur, Limita- tions of current robot-assisted pedicle screw insertion systems, Neuro- surgical Focus 57 (6) (2024) E14

    work page 2024

  10. [10]

    N. J. Lee, I. A. Buchanan, V. Boddapati, J. Mathew, G. Marciano, P. J. Park, E. Leung, A. L. Buchholz, J. Pollina, E. Jazini, et al., Do robot-related complications influence 1 year reoperations and other clin- ical outcomes after robot-assisted lumbar arthrodesis? a multicenter as- sessment of 320 patients, Journal of Orthopaedic Surgery and Research 16 ...

    work page 2021

  11. [11]

    Kubicek, F

    J. Kubicek, F. Tomanec, M. Cerny, D. Vilimek, M. Kalova, D. Oczka, Recent trends, technical concepts and components of computer-assisted 29 orthopedic surgery systems: a comprehensive review, Sensors 19 (23) (2019) 5199

    work page 2019

  12. [12]

    Zheng, L

    G. Zheng, L. P. Nolte, Computer-assisted orthopedic surgery: current state and future perspective, Frontiers in surgery 2 (2015) 66

    work page 2015

  13. [13]

    Tonetti, M

    J. Tonetti, M. Boudissa, G. Kerschbaumer, O. Seurat, Role of 3d intra- operative imaging in orthopedic and trauma surgery, Orthopaedics & Traumatology: Surgery & Research 106 (1) (2020) S19–S25

    work page 2020

  14. [14]

    S. Park, W. Plishker, H. Quon, J. Wong, R. Shekhar, J. Lee, De- formable registration of ct and cone-beam ct with local intensity match- ing, Physics in Medicine & Biology 62 (3) (2017) 927

    work page 2017

  15. [15]

    Docquier, L

    P.-L. Docquier, L. Paul, O. Cartiaux, C. Delloye, X. Banse, Computer- assisted resection and reconstruction of pelvic tumor sarcoma, Sarcoma 2010 (1) (2010) 125162

    work page 2010

  16. [16]

    Pojskić, M

    M. Pojskić, M. Bopp, B. Saß, A. Kirschbaum, C. Nimsky, B. Carl, Intraoperative computed tomography-based navigation with augmented reality for lateral approaches to the spine, Brain Sciences 11 (5) (2021) 646

    work page 2021

  17. [17]

    Markelj, D

    P. Markelj, D. Tomaževič, B. Likar, F. Pernuš, A review of 3d/2d regis- tration methods for image-guided interventions, Medical image analysis 16 (3) (2012) 642–661

    work page 2012

  18. [18]

    Otake, M

    Y. Otake, M. Armand, R. S. Armiger, M. D. Kutzer, E. Basafa, P. Kazanzides, R. H. Taylor, Intraoperative image-based multiview 2d/3d registration for image-guided orthopaedic surgery: incorporation offiducial-basedc-armtrackingandgpu-acceleration, IEEEtransactions on medical imaging 31 (4) (2011) 948–962

    work page 2011

  19. [19]

    G. P. Penney, J. Weese, J. A. Little, P. Desmedt, D. L. Hill, et al., A comparison of similarity measures for use in 2-d-3-d medical image registration, IEEE transactions on medical imaging 17 (4) (1998) 586– 595

    work page 1998

  20. [20]

    P.-C. Ku, A. Martin-Gomez, C. Gao, R. Grupp, S. C. Mears, M. Ar- mand, Towards 2d/3d registration of the preoperative mri to intraop- erative fluoroscopic images for visualisation of bone defects, Computer 30 Methods in Biomechanics and Biomedical Engineering: Imaging & Vi- sualization 11 (4) (2023) 1096–1105

    work page 2023

  21. [21]

    Hacihaliloglu, A

    I. Hacihaliloglu, A. Rasoulian, R. N. Rohling, P. Abolmaesumi, Lo- cal phase tensor features for 3-d ultrasound to statistical shape+ pose spine model registration, IEEE transactions on Medical Imaging 33 (11) (2014) 2167–2179

    work page 2014

  22. [22]

    Roche, X

    A. Roche, X. Pennec, G. Malandain, N. Ayache, Rigid registration of 3- d ultrasound with mr images: a new approach combining intensity and gradient information, IEEE transactions on medical imaging 20 (10) (2002) 1038–1049

    work page 2002

  23. [23]

    G. P. Penney, D. C. Barratt, C. S. Chan, M. Slomczykowski, T. J. Carter, P. J. Edwards, D. J. Hawkes, Cadaver validation of intensity- based ultrasound to ct registration, Medical image analysis 10 (3) (2006) 385–395

    work page 2006

  24. [24]

    Pandey, P

    P. Pandey, P. Guy, A. J. Hodgson, R. Abugharbieh, Fast and automatic bone segmentation and registration of 3d ultrasound to ct for the full pelvic anatomy: a comparative study, International Journal of Com- puter Assisted Radiology and Surgery 13 (10) (2018) 1515–1524. URLhttp://dx.doi.org/10.1007/s11548-018-1788-5

    work page doi:10.1007/s11548-018-1788-5 2018

  25. [25]

    W. Wein, A. Khamene, D.-A. Clevert, O. Kutter, N. Navab, Simula- tion and fully automatic multimodal registration of medical ultrasound, in: Medical Image Computing and Computer-Assisted Intervention– MICCAI 2007: 10th International Conference, Brisbane, Australia, Oc- tober 29-November 2, 2007, Proceedings, Part I 10, Springer, 2007, pp. 136–143

    work page 2007

  26. [26]

    D. K. Bae, S. J. Song, Computer assisted navigation in knee arthro- plasty, Clinics in orthopedic surgery 3 (4) (2011) 259–267

    work page 2011

  27. [27]

    K. C. Anderson, K. C. Buehler, D. C. Markel, Computer assisted naviga- tion in total knee arthroplasty: comparison with conventional methods, The Journal of arthroplasty 20 (2005) 132–138

    work page 2005

  28. [28]

    D. A. Wilson, C. Anglin, F. Ambellan, C. M. Grewe, A. Tack, H. Lamecker, M. Dunbar, S. Zachow, Validation of three-dimensional 31 models of the distal femur created from surgical navigation point cloud data for intraoperative and postoperative analysis of total knee arthroplasty, International Journal of Computer Assisted Radiology and Surgery 12 (12) (20...

    work page 2017

  29. [29]

    Liebmann, M

    F. Liebmann, M. von Atzigen, D. Stütz, J. Wolf, L. Zingg, D. Suter, N. A. Cavalcanti, L. Leoty, H. Esfandiari, J. G. Snedeker, M. R. Os- wald, M. Pollefeys, M. Farshad, P. Fürnstahl, Automatic registration withcontinuousposeupdatesformarker-lesssurgicalnavigationinspine surgery, Medical Image Analysis 91 (2024) 103027

    work page 2024

  30. [30]

    X. Hu, A. Nguyen, F. R. y Baena, Occlusion-robust visual markerless bone tracking for computer-assisted orthopedic surgery, IEEE Transac- tions on Instrumentation and Measurement 71 (2021) 1–11

    work page 2021

  31. [31]

    Félix, C

    I. Félix, C. Raposo, M. Antunes, P. Rodrigues, J. P. Barreto, Towards markerless computer-aided surgery combining deep segmentation and geometric pose estimation: application in total knee arthroplasty, Com- puter Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization 9 (3) (2021) 271–278

    work page 2021

  32. [32]

    X.Hu, F.R.y.Baena, Automaticbonesurfacerestorationformarkerless computer-assisted orthopaedic surgery, Chinese Journal of Mechanical Engineering 35 (1) (2022) 18

    work page 2022

  33. [33]

    Rodrigues, M

    P. Rodrigues, M. Antunes, C. Raposo, P. Marques, F. Fonseca, J. P. Barreto, Deep segmentation leverages geometric pose estimation in computer-aided total knee arthroplasty, Healthcare technology letters 6 (6) (2019) 226–230

    work page 2019

  34. [34]

    R. B. Grupp, M. Unberath, C. Gao, R. A. Hegeman, R. J. Murphy, C. P. Alexander, Y. Otake, B. A. McArthur, M. Armand, R. H. Taylor, Automatic annotation of hip anatomy in fluoroscopy for robust and efficient 2d/3d registration, International journal of computer assisted radiology and surgery 15 (5) (2020) 759–769

    work page 2020

  35. [35]

    P.-C. Ku, M. Liu, R. Grupp, A. Harris, J. K. Oni, S. C. Mears, A. Martin-Gomez, M. Armand, End-to-end 2d/3d registration from pre- operative mri to intra-operative fluoroscopy for orthopedic procedures, 32 International Journal of Computer Assisted Radiology and Surgery (2025) 1–12

    work page 2025

  36. [36]

    Leung, N

    K. Leung, N. Tang, L. Cheung, E. Ng, Image-guided navigation in or- thopaedic trauma, The Journal of Bone & Joint Surgery British Volume 92 (10) (2010) 1332–1337

    work page 2010

  37. [37]

    W. Wein, A. Karamalis, A. Baumgartner, N. Navab, Automatic bone detection and soft tissue aware ultrasound–ct registration for computer- aided orthopedic surgery, International journal of computer assisted ra- diology and surgery 10 (2015) 971–979

    work page 2015

  38. [38]

    Salehi, R

    M. Salehi, R. Prevost, J.-L. Moctezuma, N. Navab, W. Wein, Precise ul- trasound bone registration with learning-based segmentation and speed of sound calibration, paper presented at Medical Image Computing and Computer-Assisted Intervention−MICCAI (2017)

    work page 2017

  39. [39]

    K. A. Eppenhof, M. W. Lafarge, M. Veta, J. P. Pluim, Progressively trained convolutional neural networks for deformable image registration, IEEE transactions on medical imaging 39 (5) (2019) 1594–1604

    work page 2019

  40. [40]

    H. Guo, X. Xu, S. Xu, B. J. Wood, P. Yan, End-to-end ultrasound frame to volume registration, in: Medical Image Computing and Com- puter Assisted Intervention–MICCAI 2021: 24th International Confer- ence, Strasbourg, France, September 27–October 1, 2021, Proceedings, Part IV 24, Springer, 2021, pp. 56–65

    work page 2021

  41. [41]

    URLhttp://dx.doi.org/10.1007/s11548-018-1756-0

    M.Ciganovic, F.Ozdemir, F.Pean, P.Fuernstahl, C.Tanner, O.Goksel, Registration of 3d freehand ultrasound to a bone model for orthopedic procedures of the forearm, International Journal of Computer Assisted Radiology and Surgery 13 (6) (2018) 827–836. URLhttp://dx.doi.org/10.1007/s11548-018-1756-0

    work page doi:10.1007/s11548-018-1756-0 2018

  42. [42]

    Fanti, F

    Z. Fanti, F. Torres, E. Hazan-Lasri, A. Gastelum-Strozzi, L. Ruiz- Huerta, A. Caballero-Ruiz, F. A. Cosío, Improved surface-based reg- istration of ct and intraoperative 3d ultrasound of bones, Journal of healthcare engineering 2018 (1) (2018) 2365178

    work page 2018

  43. [43]

    R. Li, A. Davoodi, Y. Cai, K. Niu, G. Borghesan, N. Cavalcanti, A. Mas- salimova, F. Carrillo, C. J. Laux, M. Farshad, P. Fürnstahl, E. V. 33 Poorten, Robot-assisted ultrasound reconstruction for spine surgery: from bench-top to pre-clinical study, International Journal of Computer Assisted Radiology and Surgery 18 (9) (2023) 1613–1623. URLhttp://dx.doi.o...

    work page doi:10.1007/s11548-023-02932-z 2023

  44. [44]

    Sotiras, C

    A. Sotiras, C. Davatzikos, N. Paragios, Deformable medical image regis- tration: A survey, IEEE transactions on medical imaging 32 (7) (2013) 1153–1190

    work page 2013

  45. [45]

    J.Woo, M.Stone, J.L.Prince, Multimodalregistrationviamutualinfor- mation incorporating geometric and spatial context, IEEE Transactions on Image Processing 24 (2) (2014) 757–769

    work page 2014

  46. [46]

    R. Li, A. Davoodi, Y. Cai, G. Borghesan, N. Cavalcanti, C. J. Laux, M. Farshad, F. Carrillo, P. Fürnstahl, E. Vander Poorten, Development and evaluation of robot-assisted ultrasound navigation system for pedi- cle screw placement: An ex-vivo animal validation, The International Journal of Medical Robotics and Computer Assisted Surgery 20 (1) (10 2023). UR...

    work page doi:10.1002/rcs.2590 2023

  47. [47]

    L. Wu, N. A. Cavalcanti, M. Seibold, G. Loggia, L. Reissner, J. Hein, S. Beeler, A. Viehöfer, S. Wirth, L. Calvet, P. Fürnstahl, Ultra- bones100k: A reliable automated labeling method and large-scale dataset for ultrasound-based bone surface extraction, Computers in Bi- ology and Medicine 194 (2025) 110435

    work page 2025

  48. [48]

    J. N. Zhang, Y. Fan, D. J. Hao, Risk factors for robot-assisted spinal pedicle screw malposition, Scientific reports 9 (1) (2019) 3025

    work page 2019

  49. [49]

    L. Ma, Z. Zhao, F. Chen, B. Zhang, L. Fu, H. Liao, Augmented real- ity surgical navigation with ultrasound-assisted registration for pedicle screw placement: a pilot study, International journal of computer as- sisted radiology and surgery 12 (12) (2017) 2205–2215

    work page 2017

  50. [50]

    M. A. Fischler, R. C. Bolles, Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography, Communications of the ACM 24 (6) (1981) 381–395

    work page 1981

  51. [51]

    Q.-Y. Zhou, J. Park, V. Koltun, Fast global registration, in: European conference on computer vision, Springer, 2016, pp. 766–782. 34

    work page 2016

  52. [52]

    Gsaxner, A

    C. Gsaxner, A. Pepe, J. Wallner, D. Schmalstieg, J. Egger, Markerless image-to-face registration for untethered augmented reality in head and neck surgery, in: International Conference on Medical Image Computing and Computer-Assisted Intervention, Springer, 2019, pp. 236–244

    work page 2019

  53. [53]

    L. Liu, Y. Zhao, A. Li, X. Yu, X. Xiao, S. Liu, M. Q.-H. Meng, A photoacoustics-enhanced drilling probe for radiation-free pedicle screw implantation in spinal surgery, Frontiers in Bioengineering and Biotech- nology 10 (2022) 1000950

    work page 2022

  54. [54]

    Y. Ao, H. Esfandiari, F. Carrillo, C. J. Laux, Y. As, R. Li, K. Van Ass- che, A. Davoodi, N. A. Cavalcanti, M. Farshad, et al., Saferplan: Safe deep reinforcement learning for intraoperative planning of pedicle screw placement, Medical Image Analysis 99 (2025) 103345

    work page 2025

  55. [55]

    J. Liu, J. Hao, H. Lin, W. Pan, J. Yang, Y. Feng, G. Wang, J. Li, Z. Jin, Z. Zhao, et al., Deep learning-enabled 3d multimodal fusion of cone-beam ct and intraoral mesh scans for clinically applicable tooth- bone reconstruction, Patterns 4 (9) (2023)

    work page 2023

  56. [56]

    J. Hao, J. Liu, J. Li, W. Pan, R. Chen, H. Xiong, K. Sun, H. Lin, W. Liu, W. Ding, et al., Ai-enabled automatic multimodal fusion of cone-beam ct and intraoral scans for intelligent 3d tooth-bone reconstruction and clinical applications, arXiv preprint arXiv:2203.05784 (2022)

    work page arXiv 2022

  57. [57]

    A. Chan, B. Coutts, E. Parent, E. Lou, Development and evaluation of ct-to-3d ultrasound image registration algorithm in vertebral phantoms for spine surgery, Annals of Biomedical Engineering 49 (1) (2021) 310– 321

    work page 2021

  58. [58]

    Unberath, C

    M. Unberath, C. Gao, Y. Hu, M. Judish, R. H. Taylor, M. Armand, R. Grupp, The impact of machine learning on 2d/3d registration for image-guided interventions: A systematic review and perspective, Fron- tiers in Robotics and AI 8 (2021) 716007

    work page 2021

  59. [59]

    Zhang, J

    Y.-X. Zhang, J. Gui, X. Cong, X. Gong, W. Tao, A comprehensive survey and taxonomy on point cloud registration based on deep learning, arXiv preprint arXiv:2404.13830 (2024)

    work page arXiv 2024

  60. [60]

    U.Castellani, A.Bartoli, 3dshaperegistration, in: 3DImaging, Analysis and Applications, Springer, 2020, pp. 353–411. 35

    work page 2020

  61. [61]

    Liebmann, D

    F. Liebmann, D. Stütz, D. Suter, S. Jecklin, J. G. Snedeker, M. Farshad, P. Fürnstahl, H. Esfandiari, Spinedepth: A multi-modal data collection approach for automatic labelling and intraoperative spinal shape recon- struction based on rgb-d data, Journal of Imaging 7 (9) (2021) 164. URLhttp://dx.doi.org/10.3390/jimaging7090164

    work page doi:10.3390/jimaging7090164 2021

  62. [62]

    P. J. Besl, N. D. McKay, Method for registration of 3-d shapes, in: Sensor fusion IV: control paradigms and data structures, Vol. 1611, Spie, 1992, pp. 586–606

    work page 1992

  63. [63]

    E. W. Nottmeier, T. L. Crosby, Timing of paired points and surface matching registration in three-dimensional (3d) image-guided spinal surgery, Clinical Spine Surgery 20 (4) (2007) 268–270

    work page 2007

  64. [64]

    Foroughi, R

    P. Foroughi, R. H. Taylor, G. Fichtinger, Automatic initialization for 3d bone registration, in: Medical Imaging 2008: Visualization, Image- Guided Procedures, and Modeling, Vol. 6918, SPIE, 2008, pp. 895–902

    work page 2008

  65. [65]

    Zhang, Y

    Z. Zhang, Y. Dai, J. Sun, Deep learning based point cloud registration: an overview, Virtual Reality & Intelligent Hardware 2 (3) (2020) 222– 246

    work page 2020

  66. [66]

    C. Choy, J. Park, V. Koltun, Fully convolutional geometric features, in: Proceedings of the IEEE/CVF international conference on computer vision, 2019, pp. 8958–8966

    work page 2019

  67. [67]

    Gojcic, C

    Z. Gojcic, C. Zhou, J. D. Wegner, A. Wieser, The perfect match: 3d point cloud matching with smoothed densities, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 5545–5554

    work page 2019

  68. [68]

    S. Ao, Q. Hu, B. Yang, A. Markham, Y. Guo, Spinnet: Learning a gen- eral surface descriptor for 3d point cloud registration, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, pp. 11753–11762

    work page 2021

  69. [69]

    A. Zeng, S. Song, M. Nießner, M. Fisher, J. Xiao, T. Funkhouser, 3dmatch: Learning local geometric descriptors from rgb-d reconstruc- tions, in: Proceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 1802–1811. 36

    work page 2017

  70. [70]

    H. Wang, Y. Liu, Z. Dong, W. Wang, You only hypothesize once: Point cloud registration with rotation-equivariant descriptors, in: Proceedings of the 30th ACM International Conference on Multimedia, 2022, pp. 1630–1641

    work page 2022

  71. [71]

    Huang, Z

    S. Huang, Z. Gojcic, M. Usvyatsov, A. Wieser, K. Schindler, Predator: Registration of 3d point clouds with low overlap, in: Proceedings of the IEEE/CVF Conference on computer vision and pattern recognition, 2021, pp. 4267–4276

    work page 2021

  72. [72]

    G. D. Pais, S. Ramalingam, V. M. Govindu, J. C. Nascimento, R. Chel- lappa, P. Miraldo, 3dregnet: A deep neural network for 3d point reg- istration, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 7193–7203

    work page 2020

  73. [73]

    W. Yuan, B. Eckart, K. Kim, V. Jampani, D. Fox, J. Kautz, Deepgmr: Learning latent gaussian mixture models for registration, in: European conference on computer vision, Springer, 2020, pp. 733–750

    work page 2020

  74. [74]

    Y. Wang, J. M. Solomon, Deep closest point: Learning representations for point cloud registration, in: Proceedings of the IEEE/CVF interna- tional conference on computer vision, 2019, pp. 3523–3532

    work page 2019

  75. [75]

    W. Lu, G. Wan, Y. Zhou, X. Fu, P. Yuan, S. Song, Deepvcp: An end-to- end deep neural network for point cloud registration, in: Proceedings of the IEEE/CVF international conference on computer vision, 2019, pp. 12–21

    work page 2019

  76. [76]

    H. Xu, N. Ye, G. Liu, B. Zeng, S. Liu, Finet: Dual branches feature interaction for partial-to-partial point cloud registration, in: Proceed- ings of the AAAI Conference on Artificial Intelligence, Vol. 36, 2022, pp. 2848–2856

    work page 2022

  77. [77]

    Z. Qin, H. Yu, C. Wang, Y. Guo, Y. Peng, S. Ilic, D. Hu, K. Xu, Geo- transformer: Fast and robust point cloud registration with geometric transformer, IEEE Transactions on Pattern Analysis and Machine In- telligence 45 (8) (2023) 9806–9821

    work page 2023

  78. [78]

    H. Deng, T. Birdal, S. Ilic, Ppf-foldnet: Unsupervised learning of ro- tation invariant 3d local descriptors, in: Proceedings of the European conference on computer vision (ECCV), 2018, pp. 602–618. 37

    work page 2018

  79. [79]

    G. Mei, H. Tang, X. Huang, W. Wang, J. Liu, J. Zhang, L. Van Gool, Q. Wu, Unsupervised deep probabilistic approach for partial point cloud registration, in: ProceedingsoftheIEEE/CVFConferenceonComputer Vision and Pattern Recognition, 2023, pp. 13611–13620

    work page 2023

  80. [80]

    L. Wu, M. Seibold, N. A. Cavalcanti, G. Loggia, L. Reissner, B. Sigrist, J. Hein, L. Calvet, A. Viehöfer, P. Fürnstahl, Ultraboneudf: Self- supervised bone surface reconstruction from ultrasound based on neural unsigned distance functions, arXiv preprint arXiv:2505.17912 (2025)

    work page arXiv 2025

Showing first 80 references.