arxiv: 2604.17846 · v1 · submitted 2026-04-20 · 💻 cs.CV · cs.AI

Recognition: unknown

AI Approach for MRI-only Full-Spine Vertebral Segmentation and 3D Reconstruction in Paediatric Scoliosis

Nathasha Naranpanawa , Maree T. Izatt , Robert D. Labrom , Geoffrey N. Askin , J. Paige Little

Authors on Pith no claims yet

Pith reviewed 2026-05-10 05:16 UTC · model grok-4.3

classification 💻 cs.CV cs.AI

keywords AI segmentationMRI spine reconstructionpaediatric scoliosisGAN synthetic imagesU-Net model3D vertebral modelingradiation-free imagingthoracolumbar spine

0 comments

The pith

An AI model trained partly on synthetic MRI images from CT scans can automatically segment the full thoracolumbar spine and produce 3D reconstructions from real MRI alone.

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

The paper seeks to prove that a combination of image translation and segmentation networks can deliver accurate, continuous 3D models of the entire thoracolumbar spine directly from standard MRI scans. This would matter because MRI avoids ionizing radiation for children, yet 3D bony assessment of scoliosis has stayed manual and slow. If the method works, clinicians could obtain deformity measurements and surgical plans in under a minute while keeping the specific curves and vertebral shapes of adolescent idiopathic scoliosis intact.

Core claim

Historical low-dose CT scans from scoliosis patients are turned into MRI-style images by a generative adversarial network and mixed with existing labelled thoracic MRI to train a U-Net; the resulting model then segments vertebrae from T1 to L5 on unseen MRI scans, yielding continuous 3D reconstructions at 88 percent Dice overlap in less than one minute while preserving the original deformity geometry.

What carries the argument

The GAN-to-U-Net pipeline, in which a generative adversarial network creates realistic MRI-like images from CT data to augment training, and the U-Net then performs pixel-wise vertebral segmentation on real MRI inputs to enable the 3D output.

If this is right

Thoracolumbar 3D reconstructions become fully automated and continuous across all vertebrae.
Processing time falls from roughly one hour of manual work to under one minute per scan.
AIS-specific deformity features such as vertebral rotation and curve angles remain present in the output models.
Radiation-free 3D assessment supports clinical evaluation, surgical planning, and navigation.

Where Pith is reading between the lines

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

The same synthetic-data approach could support repeated MRI monitoring of spine growth without cumulative radiation risk.
MRI-derived models might feed directly into surgical navigation systems that currently rely on CT.
The method could be tested on other spinal pathologies or adult populations once additional labelled MRI sets become available.

Load-bearing premise

Images synthetically generated by the GAN from CT scans are close enough in appearance and distribution to real clinical MRI that the trained model generalizes without losing accuracy or deformity detail on actual patient scans.

What would settle it

Apply the trained model to a held-out set of real paediatric MRI scans, compute the Dice score for vertebral segmentation, and check whether the resulting 3D models match the true spinal curvatures and vertebral positions measured from the original images to within the reported accuracy.

read the original abstract

MRI is preferred over CT in paediatric imaging because it avoids ionising radiation, but its use in spine deformity assessment is largely limited by the lack of automated, high-resolution 3D bony reconstruction, which continues to rely on CT. MRI-based 3D reconstruction remains impractical due to manual workflows and the scarcity of labelled full-spine datasets. This study introduces an AI framework that enables fully automated thoracolumbar spine (T1-L5) segmentation and 3D reconstruction from MRI alone. Historical low-dose CT scans from adolescent idiopathic scoliosis (AIS) patients were converted into MRI-like images using a GAN and combined with existing labelled thoracic MRI data to train a U-Net-based model. The resulting algorithm accurately generated continuous thoracolumbar 3D reconstructions, improved segmentation accuracy (88% Dice score), and reduced processing time from approximately 1 hour to under one minute, while preserving AIS-specific deformity features. This approach enables radiation-free 3D deformity assessment from MRI, supporting clinical evaluation, surgical planning, and navigation in paediatric spine care.

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.

Desk Editor's Note private letter to a colleague

The paper gives a workable GAN-to-U-Net pipeline for full-spine MRI segmentation in kids with scoliosis, but the performance numbers rest on untested assumptions about how well the synthetic images match real scans.

read the letter

The main new piece is an end-to-end setup that takes historical low-dose CT scans from adolescent idiopathic scoliosis patients, runs them through a CycleGAN to make MRI-like versions, mixes those with existing labeled thoracic MRI, and trains a U-Net to segment T1-L5 from actual MRI. It then builds 3D reconstructions and claims 88% Dice plus a drop from an hour of manual work to under a minute, while keeping the curve shapes intact. That addresses a real bottleneck: MRI is safer for kids than CT, but full-spine bony segmentation has been stuck on manual methods because labeled data is scarce below the thorax. The time saving and radiation-free angle are the practical wins here. The approach is a straightforward extension of existing translation-plus-segmentation work rather than a new algorithm, but applying it specifically to full-spine pediatric deformity is a useful engineering step. The soft spot is exactly the one the stress test flags. The abstract gives the Dice score and deformity preservation without cohort sizes, cross-validation details, a held-out real full-spine test set, or any check on domain shift between the GAN outputs and clinical MRI. If the synthetic images differ in bone edge sharpness or intensity around endplates, the reported accuracy may not carry over. No ablation removing the GAN component or direct comparison to CT ground truth on the same cases is mentioned either. The math itself is standard and not circular. This is for readers who build or use automated tools in pediatric spine imaging and orthopedics. Someone already working on similar GAN-augmented segmentation would get concrete implementation ideas and see the clinical motivation clearly. It deserves peer review because the problem is important and the pipeline is reproducible in principle; referees can require the missing validation numbers before any stronger claims.

Referee Report

3 major / 2 minor

Summary. The manuscript presents an AI framework for fully automated thoracolumbar (T1-L5) vertebral segmentation and 3D reconstruction from MRI in pediatric adolescent idiopathic scoliosis (AIS) patients. Historical low-dose CT scans are converted to MRI-like images via a GAN, combined with existing labeled thoracic MRI data, and used to train a U-Net model; the resulting pipeline is reported to achieve 88% Dice score, reduce processing time from ~1 hour to under one minute, and preserve AIS deformity features.

Significance. If the validation details and domain-shift concerns are resolved, the work could enable radiation-free 3D spine assessment in children, which would be a meaningful advance for clinical evaluation, surgical planning, and navigation in pediatric scoliosis care. The core idea of leveraging GAN-augmented data to extend limited thoracic MRI labels to full-spine coverage is technically plausible and addresses a genuine clinical gap.

major comments (3)

[Abstract and Results] Abstract and Results sections: the central claim of 88% Dice score and accurate 3D reconstruction on real clinical MRI is presented without reporting the size of the validation cohort, the cross-validation strategy, whether the test set contains unseen real full-spine pediatric MRI (as opposed to synthetic or thoracic-only data), or quantitative comparison against manual segmentations or CT-derived ground truth. These omissions make it impossible to assess the reliability of the performance numbers.
[Methods] Methods section: the pipeline relies on the assumption that GAN-converted low-dose CT images are distributionally matched to real pediatric MRI, yet no quantitative domain-shift metrics (e.g., intensity histogram overlap, Fréchet Inception Distance, or edge-sharpness comparisons) or ablation removing the GAN component are provided. Without such evidence, the generalization claim to unseen real full-spine MRI cannot be evaluated.
[Results] Results section: the statement that 'AIS-specific deformity features' are preserved in the 3D reconstructions lacks supporting quantitative data such as Cobb-angle agreement, vertebral rotation metrics, or expert radiologist scoring on real-MRI test cases. This is load-bearing for the clinical utility claim.

minor comments (2)

[Abstract] Abstract: the number of historical CT scans and patients used for GAN training is not stated; this detail should be added for reproducibility.
Ensure first use of acronyms (AIS, GAN, U-Net, Dice) is accompanied by full expansion.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the detailed and insightful comments on our manuscript. We agree that additional details on validation, domain adaptation, and quantitative evaluation of clinical features are necessary to fully support our claims. Below we address each major comment and indicate the revisions we will make to the manuscript.

read point-by-point responses

Referee: [Abstract and Results] Abstract and Results sections: the central claim of 88% Dice score and accurate 3D reconstruction on real clinical MRI is presented without reporting the size of the validation cohort, the cross-validation strategy, whether the test set contains unseen real full-spine pediatric MRI (as opposed to synthetic or thoracic-only data), or quantitative comparison against manual segmentations or CT-derived ground truth. These omissions make it impossible to assess the reliability of the performance numbers.

Authors: We acknowledge the validity of this observation. The current manuscript does not report the validation cohort size, cross-validation strategy, or explicitly confirm the composition of the test set in the abstract and results sections. We will revise these sections to include the size of the validation cohort of real full-spine pediatric MRI scans, the cross-validation approach used, confirmation that the test set consists of unseen real full-spine data, and details of the quantitative comparison to manual segmentations. This will allow readers to better assess the reliability of the 88% Dice score. revision: yes
Referee: [Methods] Methods section: the pipeline relies on the assumption that GAN-converted low-dose CT images are distributionally matched to real pediatric MRI, yet no quantitative domain-shift metrics (e.g., intensity histogram overlap, Fréchet Inception Distance, or edge-sharpness comparisons) or ablation removing the GAN component are provided. Without such evidence, the generalization claim to unseen real full-spine MRI cannot be evaluated.

Authors: We agree that quantitative evidence for the distributional match between GAN-generated images and real MRI is important. The manuscript currently lacks domain-shift metrics and an ablation study without the GAN. We will add quantitative metrics such as intensity histogram overlap and Fréchet Inception Distance, along with an ablation experiment removing the GAN component, to the methods and results sections. This will support the claim of generalization to real full-spine MRI. revision: yes
Referee: [Results] Results section: the statement that 'AIS-specific deformity features' are preserved in the 3D reconstructions lacks supporting quantitative data such as Cobb-angle agreement, vertebral rotation metrics, or expert radiologist scoring on real-MRI test cases. This is load-bearing for the clinical utility claim.

Authors: We recognize that the preservation of AIS-specific deformity features requires quantitative support for the clinical utility claim. The manuscript currently relies on qualitative demonstration. We will revise the results section to include quantitative data, such as agreement in Cobb angles and vertebral rotation metrics between the automated 3D reconstructions and manual assessments on real-MRI test cases, as well as expert scoring where applicable. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical ML pipeline with independent validation

full rationale

The paper presents a standard supervised learning pipeline: GAN-based synthesis of MRI-like images from CT, combination with existing thoracic MRI labels, U-Net training, and reporting of Dice score plus qualitative deformity preservation on (presumably held-out) clinical MRI. No equations, uniqueness theorems, or first-principles derivations are invoked. The reported accuracy is an empirical measurement on test data, not a re-expression of training inputs or fitted parameters by construction. No self-citation load-bearing steps appear in the provided abstract or description.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 0 invented entities

The central claim rests on the assumption that synthetic MRI images preserve enough anatomical and intensity fidelity for downstream segmentation; no new physical entities are postulated. Neural-network weights and GAN training hyperparameters function as free parameters fitted to the combined dataset.

free parameters (2)

GAN training hyperparameters and loss weights
CycleGAN or equivalent translation network parameters chosen to map CT to MRI-like appearance; these are fitted during the image synthesis stage.
U-Net architecture and training hyperparameters
Number of layers, filter counts, learning rate, augmentation choices, and loss function weights fitted on the combined synthetic and real MRI dataset.

axioms (2)

domain assumption The distribution of vertebral shapes and intensities in the historical low-dose CT cohort is representative of the target pediatric MRI population.
Invoked when claiming that GAN outputs enable generalization to real clinical MRI.
domain assumption Dice score on held-out synthetic or limited real data is a sufficient proxy for clinical utility in 3D deformity assessment.
Used to assert that 88% Dice supports surgical planning and navigation use cases.

pith-pipeline@v0.9.0 · 5509 in / 1716 out tokens · 29606 ms · 2026-05-10T05:16:00.196734+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

33 extracted references · 30 canonical work pages · 1 internal anchor

[1]

Diagnostic Approach to Pediatric Spine Disorders

Rossi A, Martinetti C, Morana G, Severino M, Tortora D. Diagnostic Approach to Pediatric Spine Disorders. Magn Reson Imaging Clin N Am. 2016 Aug 1;24(3):621–44. doi:10.1016/j.mric.2016.04.001 PubMed PMID: 27417404

work page doi:10.1016/j.mric.2016.04.001 2016
[2]

Diagnostic Excellence in Pediatric Spine Imaging: Using Contextualized Imaging Protocols

Kadom N, Reddy K, Cooper ME, Knight-Scott J, Jones RA, Palasis S. Diagnostic Excellence in Pediatric Spine Imaging: Using Contextualized Imaging Protocols. Diagnostics 2023, V ol 13,. 2023 Sep 18;13(18). doi:10.3390/diagnostics13182973

work page doi:10.3390/diagnostics13182973 2023
[3]

Computed Tomography — An Increasing Source of Radiation Exposure

Brenner DJ, Hall EJ. Computed Tomography — An Increasing Source of Radiation Exposure. New England Journal of Medicine. 2007 Nov 29;357(22):2277–84. doi:10.1056/nejmra072149 PubMed PMID: 18046031

work page doi:10.1056/nejmra072149 2007
[4]

Radiation exposure during the treatment of spinal deformities

Mehta JS, Hodgson K, Yiping L, Kho JSB, Thimmaiah R, Topiwala U, et al. Radiation exposure during the treatment of spinal deformities. Bone Joint J. 2021 Apr 1;103–B(4):1–7. doi:10.1302/0301-620X.103B.BJJ-2020-1416.R3 PubMed PMID: 33595351

work page doi:10.1302/0301-620x.103b.bjj-2020-1416.r3 2021
[5]

Intraoperative navigation increases the projected lifetime cancer risk in patients undergoing surgery for adolescent idiopathic scoliosis

Striano BM, Crawford AM, Verhofste BP, Hresko AM, Hedequist DJ, Schoenfeld AJ, et al. Intraoperative navigation increases the projected lifetime cancer risk in patients undergoing surgery for adolescent idiopathic scoliosis. Spine J. 2024 Jun 1;24(6):1087–94. doi:10.1016/j.spinee.2024.01.007 PubMed PMID: 38262498

work page doi:10.1016/j.spinee.2024.01.007 2024
[6]

Adolescent idiopathic scoliosis: diagnosis and management

Horne JP , Flannery R, Usman S. Adolescent idiopathic scoliosis: diagnosis and management. Am Fam Physician. 2014;89(3):193–8. PubMed PMID: 24506121

2014
[7]

Magnetic resonance imaging effectiveness in adolescent idiopathic scoliosis

de Oliveira RG, de Araújo AO, Gomes CR. Magnetic resonance imaging effectiveness in adolescent idiopathic scoliosis. Spine Deform. 2020 Sep 17;9(1):67–73. doi:10.1007/s43390- 020-00205-2 PubMed PMID: 32940878

work page doi:10.1007/s43390- 2020
[8]

Clinical Value of Routine Preoperative Magnetic Resonance Imaging in Adolescent Idiopathic Scoliosis A PROSPECTIVE STUDY OF THREE HUNDRED AND TWENTY-SEVEN PATIENTS

Do T, Burke S, Widmann RF, Rawlins B, Boachie-Adjei O. Clinical Value of Routine Preoperative Magnetic Resonance Imaging in Adolescent Idiopathic Scoliosis A PROSPECTIVE STUDY OF THREE HUNDRED AND TWENTY-SEVEN PATIENTS. V ol
[9]

Indications for Magnetic Resonance Imaging in Presumed Adolescent Idiopathic Scoliosis

Davids JR, Chamberlin E, Blackhurst DW. Indications for Magnetic Resonance Imaging in Presumed Adolescent Idiopathic Scoliosis. THE JOURNAL OF BONE AND JOINT SURGERY . 2004. Report

2004
[10]

Indication for preoperative MRI of neural axis abnormalities in patients with presumed thoracolumbar/lumbar idiopathic scoliosis

Qiao J, Zhu Z, Zhu F, Wu T, Qian B, Xu L, et al. Indication for preoperative MRI of neural axis abnormalities in patients with presumed thoracolumbar/lumbar idiopathic scoliosis. European Spine Journal. 2013 Feb;22(2):360–6. doi:10.1007/s00586-012-2557-8 PubMed PMID: 23143092

work page doi:10.1007/s00586-012-2557-8 2013
[11]

Garg S, Darland H, Kim E, Sanchez B, Carry P. 7.7% Prevalence of neural axis abnormalities on routine magnetic resonance imaging in patients with presumed adolescent idiopathic scoliosis scheduled for spine surgery: a consecutive single surgeon retrospective cohort of 182 patients. Spine Deform. 2023 Jan 1;11(1):95–104. doi:10.1007/s43390-022-00568-8 PubM...

work page doi:10.1007/s43390-022-00568-8 2023
[12]

The use of physical biomodelling in complex spinal surgery

Izatt MT, Thorpe PLPJ, Thompson RG, D’Urso PS, Adam CJ, Earwaker JWS, et al. The use of physical biomodelling in complex spinal surgery. European Spine Journal 2007 16:9. 2007 Feb 14;16(9):1507–18. doi:10.1007/s00586-006-0289-3 PubMed PMID: 17846803

work page doi:10.1007/s00586-006-0289-3 2007
[13]

Biomodeling as an aid to spinal instrumentation

D’Urso PS, Williamson OD, Thompson RG. Biomodeling as an aid to spinal instrumentation. Spine (Phila Pa 1976). 2005 Dec;30(24):2841–5. doi:10.1097/01.brs.0000190886.56895.3d PubMed PMID: 16371915

work page doi:10.1097/01.brs.0000190886.56895.3d 1976
[14]

Deep Learning- Based Automatic Segmentation for Reconstructing Vertebral Anatomy of Healthy Adolescents and Patients With Adolescent Idiopathic Scoliosis (AIS) Using MRI Data

Antico M, Little JP, Jennings H, Askin G, Labrom RD, Fontanarosa D, et al. Deep Learning- Based Automatic Segmentation for Reconstructing Vertebral Anatomy of Healthy Adolescents and Patients With Adolescent Idiopathic Scoliosis (AIS) Using MRI Data. IEEE Access. 2021;9:86811–23. doi:10.1109/ACCESS.2021.3084949

work page doi:10.1109/access.2021.3084949 2021
[15]

Parameterization of multiple Bragg curves for scanning proton beams using simultaneous fitting of multiple curves

Hoad CL, Martel AL, Kerslake R, Grevitt M. A 3D MRI sequence for computer assisted surgery of the lumbarspine. Phys Med Biol. 2001 Jul 19;46(8):N213. doi:10.1088/0031- 9155/46/8/403 PubMed PMID: 11512626

work page doi:10.1088/0031- 2001
[16]

Automated detection, 3D segmentation and analysis of high resolution spine MR images using statistical shape models

Neubert A, Fripp J, Engstrom C, Schwarz R, Lauer L, Salvado O, et al. Automated detection, 3D segmentation and analysis of high resolution spine MR images using statistical shape models. Phys Med Biol. 2012 Dec 21;57(24):8357–76. doi:10.1088/0031-9155/57/24/8357 PubMed PMID: 23201861

work page doi:10.1088/0031-9155/57/24/8357 2012
[17]

Evaluation of deep learning reconstructed high-resolution 3D lumbar spine MRI

Sun S, Tan ET, Mintz DN, Sahr M, Endo Y , Nguyen J, et al. Evaluation of deep learning reconstructed high-resolution 3D lumbar spine MRI. European Radiology 2022 32:9. 2022 Mar 24;32(9):6167–77. doi:10.1007/s00330-022-08708-4 PubMed PMID: 35322280

work page doi:10.1007/s00330-022-08708-4 2022
[18]

Lumbar spine segmentation in MR images: a dataset and a public benchmark

van der Graaf JW, van Hooff ML, Buckens CFM, Rutten M, van Susante JLC, Kroeze RJ, et al. Lumbar spine segmentation in MR images: a dataset and a public benchmark. Sci Data. 2024 Dec 1;11(1). doi:10.1038/s41597-024-03090-w PubMed PMID: 38431692

work page doi:10.1038/s41597-024-03090-w 2024
[19]

Detection and Labeling of Vertebrae in MR Images Using Deep Learning with Clinical Annotations as Training Data

Forsberg D, Sjöblom E, Sunshine JL. Detection and Labeling of Vertebrae in MR Images Using Deep Learning with Clinical Annotations as Training Data. J Digit Imaging. 2017 Aug 1;30(4):406–12. doi:10.1007/s10278-017-9945-x PubMed PMID: 28083827

work page doi:10.1007/s10278-017-9945-x 2017
[20]

Assessing progressive changes in axial plane vertebral deformity in adolescent idiopathic scoliosis using sequential magnetic resonance imaging

Sowula PT, Izatt MT, Labrom RD, Askin GN, Little JP. Assessing progressive changes in axial plane vertebral deformity in adolescent idiopathic scoliosis using sequential magnetic resonance imaging. European Spine Journal 2023 33:2. 2023 Nov 14;33(2):663–72. doi:10.1007/s00586-023-08004-9 PubMed PMID: 37962687

work page doi:10.1007/s00586-023-08004-9 2023
[21]

Sequential MRI reveals vertebral body wedging significantly contributes to coronal plane deformity progression in adolescent idiopathic scoliosis during growth

Labrom FR, Izatt MT, Contractor P, Grant CA, Pivonka P, Askin GN, et al. Sequential MRI reveals vertebral body wedging significantly contributes to coronal plane deformity progression in adolescent idiopathic scoliosis during growth. Spine Deformity 2020 8:5. 2020 May 25;8(5):901–10. doi:10.1007/s43390-020-00138-w PubMed PMID: 32451976

work page doi:10.1007/s43390-020-00138-w 2020
[22]

Domain Shift in Computer Vision models for MRI data analysis: An Overview [Internet]

Kondrateva E, Pominova M, Popova E, Sharaev M, Bernstein A, Burnaev E. Domain Shift in Computer Vision models for MRI data analysis: An Overview [Internet]. 2020 Oct 14. Available from: http://arxiv.org/abs/2010.07222

work page arXiv 2020
[23]

arXiv preprint arXiv:2409.04368 (2024) 10 Dahal et al

Guo B, Lu D, Szumel G, Gui R, Wang T, Konz N, et al. The Impact of Scanner Domain Shift on Deep Learning Performance in Medical Imaging: an Experimental Study [Internet]. 2024 Oct 2. Available from: http://arxiv.org/abs/2409.04368

work page arXiv 2024
[24]

Sequential Magnetic Resonance Imaging Reveals Individual Level Deformities of Vertebrae and Discs in the Growing Scoliotic Spine

Keenan BE, Izatt MT, Askin GN, Labrom RD, Bennett DD, Pearcy MJ, et al. Sequential Magnetic Resonance Imaging Reveals Individual Level Deformities of Vertebrae and Discs in the Growing Scoliotic Spine. Spine Deform. 2017 May 1;5(3):197–207. doi:10.1016/j.jspd.2016.10.002 PubMed PMID: 28449963

work page doi:10.1016/j.jspd.2016.10.002 2017
[25]

TotalSegmentator: Robust Segmentation of 104 Anatomic Structures in CT Images

Wasserthal J, Breit HC, Meyer MT, Pradella M, Hinck D, Sauter AW, et al. TotalSegmentator: Robust Segmentation of 104 Anatomic Structures in CT Images. https://doi.org/101148/ryai230024. 2023 Jul 5;5(5). doi:10.1148/ryai.230024 PubMed PMID: 37795137

work page doi:10.1148/ryai.230024 2023
[26]

Generative adversarial networks,

Goodfellow I, Pouget-Abadie J, Mirza M, Xu B, Warde-Farley D, Ozair S, et al. Generative adversarial networks. Commun ACM. 2020 Oct 22;63(11):139–44. doi:10.1145/3422622

work page doi:10.1145/3422622 2020
[27]

U-Net: Convolutional Networks for Biomedical Image Segmentation

Ronneberger O, Fischer P, Brox T. U-Net: Convolutional Networks for Biomedical Image Segmentation [Internet]. 2015 May 18. Available from: http://arxiv.org/abs/1505.04597

work page internal anchor Pith review Pith/arXiv arXiv 2015
[28]

The safety and accuracy of radiation-free spinal navigation using a short, scoliosis- specific BoneMRI-protocol, compared to CT

Lafranca PPG, Rommelspacher Y , Walter SG, Muijs SPJ, van der Velden TA, Shcherbakova YM, et al. The safety and accuracy of radiation-free spinal navigation using a short, scoliosis- specific BoneMRI-protocol, compared to CT. European Spine Journal 2025 35:1. 2025 Jul 21;35(1):298–304. doi:10.1007/s00586-025-09151-x

work page doi:10.1007/s00586-025-09151-x 2025
[29]

Computer-Assisted Orthopedic and Trauma Surgery

Stübig T, Windhagen H, Krettek C, Ettinger M. Computer-Assisted Orthopedic and Trauma Surgery. Dtsch Arztebl Int. 2020 Nov 20;117(47):793. doi:10.3238/arztebl.2020.0793 PubMed PMID: 33549155

work page doi:10.3238/arztebl.2020.0793 2020
[30]

Robotics and navigation in spine surgery: A narrative review

Zawar A, Chhabra HS, Mundra A, Sharma S, Kalidindi KKV . Robotics and navigation in spine surgery: A narrative review. J Orthop. 2023 Oct 1;44:36. doi:10.1016/j.jor.2023.08.007 PubMed PMID: 37664556

work page doi:10.1016/j.jor.2023.08.007 2023
[31]

and Ito K, et al

Kok J, Shcherbakova YM, Schlosser TPC, Seevinck PR, van der Velden TA, Castelein Rene M. and Ito K, et al. Automatic generation of subject-specific finite element models of the spine from magnetic resonance images. Front Bioeng Biotechnol. 2023 Sep;11. doi:10.3389/fbioe.2023.1244291

work page doi:10.3389/fbioe.2023.1244291 2023
[32]

Assessment of pedicle screw placement accuracy, procedure time, and radiation exposure using a miniature robotic guidance system

Lieberman IH, Hardenbrook MA, Wang JC, Guyer RD. Assessment of pedicle screw placement accuracy, procedure time, and radiation exposure using a miniature robotic guidance system. J Spinal Disord Tech. 2012 Jul;25(5):241–8. doi:10.1097/BSD.0b013e318218a5ef PubMed PMID: 21602728

work page doi:10.1097/bsd.0b013e318218a5ef 2012
[33]

Does image guidance decrease pedicle screw-related complications in surgical treatment of adolescent idiopathic scoliosis: a systematic review update and meta-analysis

Chan A, Parent E, Wong J, Narvacan K, San C, Lou E. Does image guidance decrease pedicle screw-related complications in surgical treatment of adolescent idiopathic scoliosis: a systematic review update and meta-analysis. Eur Spine J. 2020 Apr 1;29(4):694–716. doi:10.1007/s00586-019-06219-3 PubMed PMID: 31781863

work page doi:10.1007/s00586-019-06219-3 2020