Automatic Annotation of Hip Anatomy in Fluoroscopy for Robust and Efficient 2D/3D Registration
Open accessAbstract
Fluoroscopy is the standard imaging modality used to guide hip surgery and is therefore a natural sensor for computer-assisted navigation. In order to efficiently solve the complex registration problems presented during navigation, human-assisted annotations of the intraoperative image are typically required. This manual initialization interferes with the surgical workflow and diminishes any advantages gained from navigation. We propose a method for fully automatic registration using annotations produced by a neural network. Neural networks are trained to simultaneously segment anatomy and identify landmarks in fluoroscopy. Training data is obtained using an intraoperatively incompatible 2D/3D registration of hip anatomy. Ground truth 2D labels are established using projected 3D annotations. Intraoperative registration couples an intensity-based strategy with annotations inferred by the network and requires no human assistance. Ground truth labels were obtained in 366 fluoroscopic images across 6 cadaveric specimens. In a leave-one-subject-out experiment, networks obtained mean dice coefficients for left and right hemipelves, left and right femurs of 0.86, 0.87, 0.90, and 0.84. The mean 2D landmark error was 5.0 mm. The pelvis was registered within 1 degree for 86% of the images when using the proposed intraoperative approach with an average runtime of 7 seconds. In comparison, an intensity-only approach without manual initialization, registered the pelvis to 1 degree in 18% of images. We have created the first accurately annotated, non-synthetic, dataset of hip fluoroscopy. By using these annotations as training data for neural networks, state of the art performance in fluoroscopic segmentation and landmark localization was achieved. Integrating these annotations allows for a robust, fully automatic, and efficient intraoperative registration during fluoroscopic navigation of the hip.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
7 of 12 cited sources checked · 2 supported · 1 flagged · accuracy not yet measured
- Premisecitation
Pelvis poses estimated from Bier's annotations produced reprojection errors of 14-34 mm for other landmarks not learned by the network.
“Pelvis poses were estimated using these annotations, yielding reprojection errors of 14-34 mm for other landmarks not learned by the network.”
Quote in the paperSubstantial claim1 of 1 sources checked- cites [19] Bier B, Goldmann F, Zaech JN, Fotouhi J, Hegeman R, Grupp R, Armand M, Osgood G, Navab N, Maier A, Unberath M (2019) Learning to detect anatomical landmarks of the pelvis in X-rays from arbitrary views. Int J Comput Assist Radiol Surg 14(9):1463–1473 source
The Materials and methods section (‘2D/3D registration’; page unavailable) says detected pelvis landmarks were matched to CT labels to estimate a projection matrix, supporting the pose-estimation part. The Clinical data results (page unavailable) do not report 14–34 mm reprojection errors for landmarks the network did not learn; their stated 2–5 pixel error concerns a different, metal-bead-based reference projection. (LLM, conf=0.87)
Experimental — accuracy not yet measured
- Premisecitation
Applications of CNNs to fluoroscopy have been mostly limited to recognizing surgical instruments and tools.
“However, since existing large-scale hip datasets have focused on 3D image modalities and pre and postoperative radiography, rather than intraoperative fluoroscopy [12], applications of CNNs to fluoroscopy have been mostly limited to recognizing surgical instruments and tools [13-16].”
Quote in the paperSubstantial claim2 of 4 sources checked- cites [13] Miao S, Wang ZJ, Liao R (2016) A CNN regression approach for real-time 2D/3D registration. IEEE Trans Med Imag 35(5):1352–1363 source
- cites [14] Ambrosini P, Ruijters D, Niessen WJ, Moelker A, van Walsum T (2017) Fully automatic and real-time catheter segmentation in X-ray fluoroscopy. In: Proceedings of medical image computing and computer-assisted intervention. Springer, pp 577–585 source
- cites [15] Breininger K, Albarqouni S, Kurzendorfer T, Pfister M, Kowarschik M, Maier A (2018) Intraoperative stent segmentation in X-ray fluoroscopy for endovascular aortic repair. Int J Comput Assist Radiol Surg 13(8):1221–1231 source
- cites [16] Gao C, Unberath M, Taylor R, Armand M (2019) Localizing dexterous surgical tools in X-ray for image-based navigation. arXiv preprint arXiv:1901.06672 source
The Methods and Conclusions describe a ConvNet that localizes and segments a surgical tool in X-ray images, supporting this as an example of tool-focused work. The paper does not establish that CNN applications to fluoroscopy have been *mostly limited* to instruments and tools. (LLM, conf=0.82)
Experimental — accuracy not yet measured
- Findingown data
The average landmark 2D localization error was 5.0 ± 5.2 mm in the detector plane.
“The average landmark 2D localization error was 5.0 ± 5.2 mm in the detector plane.”
Quote in the paperSubstantial claim - Findingown data
Mean segmentation and landmark detection time per image was 24.0 ± 0.4 milliseconds on a NVIDIA Tesla P100 GPU.
“The mean time for segmentation and landmark detection per image was 24.0 ± 0.4 milliseconds using a NVIDIA Tesla P100 (PCIe) GPU.”
Quote in the paperSubstantial claim - Findingown data
Mean 2D landmark localization error reached 5.0 mm.
“The mean 2D landmark localization error was 5.0 mm.”
Quote in the paperSubstantial claim
- Findingown data
Method 2 was successful with these numbers of detections in only 0, 7, and 26 cases, respectively.
“Highlighting the robustness gained from mixing intensity-features with landmark features, method 2 was only successful with these number of detections in 0, 7, and 26 cases, respectively.”
Quote in the paperSubstantial claim - Findingown data
Ground-truth segmentation labels and anatomical landmarks were acquired from 366 fluoroscopic images of six cadaveric specimens.
“Ground truth segmentation labels and anatomical landmarks were obtained in 366 fluoroscopic images across 6 cadaveric specimens.”
Quote in the paperSubstantial claim - Findingown data
Networks trained on this dataset achieved mean dice coefficients of 0.86, 0.87, 0.90, and 0.84 for the left hemipelvis, right hemipelvis, left femur, and right femur, respectively.
“In a leave-one-subject-out experiment, networks trained on this data obtained mean dice coefficients for left and right hemipelves, left and right femurs of 0.86, 0.87, 0.90, and 0.84, respectively.”
Quote in the paperSubstantial claim - Findingown data
Mean runtime for method 3 was 7.2 ± 0.7 seconds on a NVIDIA Tesla P100 GPU.
“The mean runtime for method 3 was 7.2 ± 0.7 seconds using a NVIDIA Tesla P100 (PCIe) GPU.”
Quote in the paperNot a substantial claim - Findingown data
An intensity-only method without manual initialization registered the pelvis to 1° in 18% of images.
“In comparison, an intensity-only approach without manual initialization, registered the pelvis to 1° in 18% of images.”
Quote in the paperSubstantial claim - Findingown data
Table 2 shows that method 3's mean and standard deviation of pelvis translation errors were both larger than method 2's by approximately 0.5 mm.
“Table 2 shows that the mean and standard deviation of method 3’s pelvis translation errors were both larger than those of method 2 by approximately 0.5 mm.”
Quote in the paperSubstantial claim - Findingown data
The geometries associated with specimen 6 were clearly collected at different C-arm poses than those used to train the network tested on specimen 6.
“The geometries associated with spec. 6 are clearly collected at different C-arm poses than those used for training the network tested on spec. 6.”
Quote in the paperSubstantial claim - Interpretationnone given
The paper proposes a 2D/3D registration method for hip anatomy that combines image intensities with higher-level landmark and segmentation features, making it robust against large initial offsets from actual object poses.
“In this paper, we propose a method for 2D/3D registration of hip anatomy that simultaneously combines image intensities with higher-level landmark and segmentation features, making it robust against large initial offsets from actual object poses.”
Quote in the paperSubstantial claim- Rests on an assumption: Large errors in pose estimates may occur when a registration is not initialized sufficiently close to the actual pose of an object.
- Interpretationreasoning
The larger errors of methods 2 and 3 in the remaining cases are partly caused by the more challenging pelvis registration problems in those images, which the naive method could not solve.
“Moreover, the larger errors of methods 2 and 3 in the remaining cases are in part caused by the more challenging pelvis registration problems presented in these images, for which the naive method was unable to solve.”
Quote in the paperNot a substantial claim - Interpretationown data
The three methods perform comparably on images for which the naive approach succeeds.
“This indicates that the three methods perform comparably on images for which the naive approach succeeds.”
Quote in the paperNot a substantial claim - Interpretationown data
Using these annotations as training data for neural networks yielded state-of-the-art performance in fluoroscopic segmentation and landmark localization.
“By using these annotations as training data for neural networks, state-of-the-art performance in fluoroscopic segmentation and landmark localization was achieved.”
Quote in the paperSubstantial claim - Interpretationown data
Integrating these annotations enables robust, fully automatic, and efficient intraoperative registration during fluoroscopic navigation of the hip.
“Integrating these annotations allows for a robust, fully automatic, and efficient intraoperative registration during fluoroscopic navigation of the hip.”
Quote in the paperSubstantial claim- Rests on an assumption: Human-assisted annotations of the intraoperative image are typically required to efficiently solve complex registration problems during navigation.
- Rests on an assumption: Manual initialization interferes with the surgical workflow and diminishes advantages gained from navigation.
- Interpretationnone given
The authors created the first accurately annotated, non-synthetic dataset of hip fluoroscopy.
“We have created the first accurately annotated, non-synthetic, dataset of hip fluoroscopy.”
Quote in the paperSubstantial claim - Interpretationreasoning
The registration can minimize the image similarity term provided that known 3D landmarks project to approximately the correct 2D location.
“The registration is free to minimize the image similarity term on the condition that known 3D landmarks project to approximately the correct location in 2D.”
Quote in the paperSubstantial claim - Interpretationreasoning
Restricting candidate landmark locations to pixels labeled as the corresponding anatomy avoids false alarms when the ipsilateral landmark is not in view and a large heatmap intensity appears about the contralateral landmark.
“Restricting candidate locations in this way avoids possible false alarms when the ipsilateral landmark is not in the view and a large heatmap intensity is located about the contralateral landmark.”
Quote in the paperSubstantial claim - Interpretationreasoning
The computational cost of the offline registration precludes its intraoperative application.
“It should be emphasized that, although this registration is automatic and global, the amount of computation precludes it from intraoperative application.”
Quote in the paperNot a substantial claim - Interpretationown data
The paper demonstrates that CNN models trained on small datasets of fewer than 400 annotated images can achieve state-of-the-art results for semantic segmentation and landmark localization in actual hip fluoroscopy.
“A demonstration that CNN models, trained using small datasets of less than 400 annotated images, can achieve state-of-the-art-results for the tasks of semantic segmentation and landmark localization in actual hip fluoroscopy,”
Quote in the paperSubstantial claim - Interpretationreasoning
Conversely, the registration can minimize landmark reprojection error as long as candidate poses produce images that approximately match the observed image.
“Conversely, the registration is free to minimize landmark reprojection error, so long as the candidate poses produce images that approximately match the observed image.”
Quote in the paperSubstantial claim - Premisecitation
Existing large-scale hip datasets have focused on 3D image modalities and pre- and postoperative radiography, not intraoperative fluoroscopy.
“However, since existing large-scale hip datasets have focused on 3D image modalities and pre and postoperative radiography, rather than intraoperative fluoroscopy [12], applications of CNNs to fluoroscopy have been mostly limited to recognizing surgical instruments and tools [13-16].”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [12] Otake Y, Takao M, Yokota F, Fukuda N, Uemura K, Sugano N, Sato Y (2018) Construction and application of large-scale image database in orthopedic surgery. In: Computer assisted orthopaedic surgery for hip and knee. Springer, pp 191–197 source
- Interpretationnone given
Unlike method 2, method 3's objective function never penalizes offsets of poses from their initial estimates.
“In contrast to method 2, the objective function of method 3 never places penalties on the offsets of poses from their initial estimates.”
Quote in the paperSubstantial claim - Interpretationnone given
The paper's novel contributions include a semi-automatic, offline pipeline for creating the first annotated training dataset of semantically segmented individual bone structures and anatomical landmark locations in actual hip fluoroscopy.
“A semi-automatic, offline, pipeline for creating the first annotated training dataset of semantically segmented individual bone structures and anatomical landmark locations in actual hip fluoroscopy,”
Quote in the paperSubstantial claim - Interpretationown data
The approach's performance degrades when images collected at C-arm poses not found in the training data set are processed.
“The performance of our approach degrades as images collected with C-arm poses not found in the training data set are processed.”
Quote in the paperSubstantial claim - Interpretationnone given
To the authors' knowledge, the dataset presented in this paper is the first annotated dataset of actual hip fluoroscopy, consisting of individual bone segmentations and anatomical landmark locations.
“To our knowledge, the dataset presented in this paper is the first annotated dataset of actual hip fluoroscopy, consisting of individual bone segmentations and anatomical landmark locations.”
Quote in the paperNot a substantial claim - Interpretationown data
Coupling the automatic annotations produced by the CNN models with the image intensities used during 2D/3D registrations makes robustness against poor initializations possible.
“Furthermore, we have shown that by coupling the automatic annotations produced by the CNN models with the image intensities used during 2D/3D registrations, robustness against poor initializations is possible.”
Quote in the paperSubstantial claim - Interpretationreasoning
These small segmentation errors are not expected to negatively impact registration performance because the estimated labels weight the contribution of local, overlapping patches to the image similarity term.
“These small segmentation errors are not expected to negatively impact registration performance, as the estimated labels are used to weight the contribution of local, overlapping, patches to the image similarity term.”
Quote in the paperSubstantial claim - Interpretationreasoning
A weak correlation between dice scores and registration errors was indicated by the Spearman rank coefficient values.
“Therefore, it is not surprising that a weak correlation between dice scores and registration errors was indicated by the Spearman rank coefficient values.”
Quote in the paperSubstantial claim - Interpretationreasoning
This differs from the standard regularization approach used by method 2, which limits rigid movement even when initial estimates are far from the true poses.
“This is contrary to the standard approach for regularization used by method 2, which imposes limits on the amount of rigid movement, even when the initial estimates are far from the true poses.”
Quote in the paperSubstantial claim - Interpretationnone given
Creation of the precise labels in the training dataset was made possible by extending existing 2D/3D registration technology into a new offline and semi-automatic annotation pipeline.
“Creation of the precise labels found in the training dataset was made possible by extending existing 2D/3D registration technology into a new offline and semi-automatic annotation pipeline.”
Quote in the paperSubstantial claim - Interpretationown data
In conclusion, small annotated datasets of actual hip fluoroscopy can be used to train CNN models capable of state-of-the-art segmentation and landmark localization results.
“In conclusion, this paper has demonstrated that small annotated datasets of actual hip fluoroscopy may be used to train CNN models capable of state-of-the-art segmentation and landmark localization results.”
Quote in the paperSubstantial claim
- May combine two claims. 20
- Classified as probably not a substantial claim. 4
May combine two claims.
Interpretationown dataUnlike [16,19,20], the authors train networks on smaller datasets of actual fluoroscopy and achieve state-of-the-art results on clinically relevant data.
“Contrary to [16,19,20], we train our networks using smaller datasets of actual fluoroscopy and achieve state-of-the-art results on clinically relevant data.”
Quote in the paperSubstantial claim3 of 3 sources checked- cites [16] Gao C, Unberath M, Taylor R, Armand M (2019) Localizing dexterous surgical tools in X-ray for image-based navigation. arXiv preprint arXiv:1901.06672 source
- cites [19] Bier B, Goldmann F, Zaech JN, Fotouhi J, Hegeman R, Grupp R, Armand M, Osgood G, Navab N, Maier A, Unberath M (2019) Learning to detect anatomical landmarks of the pelvis in X-rays from arbitrary views. Int J Comput Assist Radiol Surg 14(9):1463–1473 source
Bier et al. (2019) say they trained on synthetic X-rays and then applied the model to real X-rays, contrary to the claim that they trained on actual fluoroscopy. The quoted sentence appears instead in “Automatic Annotation of Hip Anatomy in Fluoroscopy for Robust and Efficient 2D/3D Registration” (unscoped score 0.52). That paper explicitly contrasts its approach with Bier’s, which it cites as [19]; it does not attribute its own training method to Bier. (LLM, conf=0.58)
Checked by a reasoning model — may be wrong.
Experimental — accuracy not yet measured - cites [20] Esteban J, Grimm M, Unberath M, Zahnd G, Navab N (2019) Towards fully automatic X-ray to CT registration. In: Proceedings of medical image computing and computer-assisted intervention. Springer, pp 631–639 source
Two searches of the cited paper returned no snippets, so its full text could not be checked. An unscoped search found the sentence nearly verbatim in “Automatic Annotation of Hip Anatomy in Fluoroscopy for Robust and Efficient 2D/3D Registration” (score 0.52), a distinct but closely related paper. That is a strong reason to question the attribution, but the match falls below the rubric’s 0.70 threshold for misattribution, and the cited paper’s abstract was unavailable for comparison. (LLM, conf=0.62)
Experimental — accuracy not yet measured
May combine two claims.
PremisecitationBier trained CNNs to annotate pelvic anatomical landmarks using simulated fluoroscopy; on five sequences of actual fluoroscopy from two cadaveric specimens, mean annotation errors of 12-24 mm in the detector plane were reported.
“Using a large collection of simulated fluoroscopy, Bier trained CNNs to annotate anatomical landmarks of the pelvis [19]. When evaluated on five sequences of actual fluoroscopy across two cadaveric specimens, mean annotation errors of 12-24 mm in the detector plane were reported.”
Quote in the paperNot a substantial claim1 of 1 sources checked- cites [19] Bier B, Goldmann F, Zaech JN, Fotouhi J, Hegeman R, Grupp R, Armand M, Osgood G, Navab N, Maier A, Unberath M (2019) Learning to detect anatomical landmarks of the pelvis in X-rays from arbitrary views. Int J Comput Assist Radiol Surg 14(9):1463–1473 source
Split into 2 atoms — 1 confirmed, 1 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.
Experimental — accuracy not yet measured
May combine two claims.
Findingown dataMethod 3's performance was robust when only 2, 3, and 4 landmarks were detected, reporting success in 2, 15, and 30 cases, respectively.
“Method 3’s performance was robust when only 2, 3, and 4 landmarks were detected; reporting success in 2, 15, and 30 cases, respectively.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataDepth estimation inaccuracies accounted for nearly all pelvis translation error, with mean errors about the X, Y, and Z axes of 0.1 ± 0.1 mm, 0.1 ± 0.1 mm, and 1.4 ± 2.0 mm, respectively, for method 3.
“Depth estimation inaccuracies accounted for nearly all of the pelvis translation error, with mean errors about the X, Y, and Z axes of 0.1 ± 0.1 mm, 0.1 ± 0.1 mm and 1.4 ± 2.0 mm, respectively for method 3.”
Quote in the paperNot a substantial claimMay combine two claims.
Findingown dataAlthough the naive registration succeeded in only 66 cases, its mean femur rotation errors were about 1° smaller than those of methods 2 and 3.
“Although the naive registration only succeeded in 66 cases, the mean femur rotation errors were about 1° smaller than those of methods 2 and 3.”
Quote in the paperNot a substantial claimMay combine two claims.
Findingown dataThe proposed intraoperative approach registered the pelvis within 1° in 86% of images, with an average runtime of 7 seconds.
“The pelvis was registered within 1° for 86% of the images when using the proposed intraoperative approach with an average runtime of 7 seconds.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataAcross all test images, mean dice coefficients were 0.86 ± 0.20, 0.87 ± 0.18, 0.90 ± 0.24, 0.84 ± 0.31, 0.74 ± 0.19, and 0.63 ± 0.13 for the left hemipelvis, right hemipelvis, left femur, right femur, vertebra, and upper sacrum, respectively.
“Across all test images, mean dice coefficients of 0.86 ± 0.20, 0.87 ± 0.18, 0.90 ± 0.24, 0.84 ± 0.31, 0.74 ± 0.19, and 0.63 ± 0.13 were obtained for the left hemipelvis, right hemipelvis, left femur, right femur, vertebra, and upper sacrum, respectively.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataMethods 2 and 3 were also successful in 62 and 64 of method 1's successes, each with a mean femur rotation error of 0.8° ± 0.5°.
“However, methods 2 and 3 were also successful in 62 and 64 of method l’s successes, each with a mean femur rotation error of 0.8° ± 0.5°.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataWith a 0.005 threshold, a significant difference was found between pelvis rotation errors of methods 2 and 3 (p < 0.001), while no significant differences were found for pelvis translation errors (p = 0.045), femur rotation errors (p = 0.089), or femur translation errors (p = 0.268).
“Using a 0.005 threshold, a significant difference was found between the pelvis rotation errors (p < 0.001), while no significant differences were found between pelvis translation errors (p = 0.045), femur rotation errors (p = 0.089), and femur translation errors (p = 0.268).”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataFor method 2, correlation coefficients between dice scores and pelvis rotation and translation errors were −0.32 and −0.29, respectively; for method 3, they were −0.31 and −0.33.
“For method 2, the correlation coefficients for the pelvis rotation and translation errors were −0.32 and −0.29, respectively, and −0.31 and −0.33, respectively for method 3.”
Quote in the paperNot a substantial claimMay combine two claims.
Findingown dataWhen testing on specimen 6, average landmark localization error was 10.0 mm, false negative rate was 30%, and successful pelvis registration rate was 29% for both methods 2 and 3.
“When testing on spec. 6, average landmark localization error was 10.0 mm, with a false negative rate of 30%, and successful pelvis registration rate of 29% for both methods 2 and 3.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationCNNs have excelled at landmark detection and semantic segmentation when sufficiently large annotated training datasets are available.
“Convolutional neural networks (CNNs) have excelled at detecting landmarks and performing semantic segmentation when sufficiently large annotated datasets are available for training [10,11].”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [10] Newell A, Yang K, Deng J (2016) Stacked hourglass networks for human pose estimation. In: Proceedings of European conference on computer vision. Springer, pp 483–499 source
- cites [11] Shelhamer E, Long J, Darrell T (2017) Fully convolutional networks for semantic segmentation. IEEE Trans Pattern Anal Mach Intell 4:640–651 source
May combine two claims.
Interpretationnone givenThe paper's online intraoperative registration strategy, leveraging image intensities and CNN features, is fully automatic, requires no user initialization, and completes in seconds.
“An online, intraoperative, registration strategy, leveraging image intensities and CNN-features, that is fully automatic, requires no initialization from a user, and completes in an order of seconds.”
Quote in the paperSubstantial claimMay combine two claims.
Interpretationown dataThe naive intraoperative registration performed poorly, succeeding in only 18% of trials, while methods leveraging CNN annotations succeeded more than four times as often.
“The naive intraoperative registration performed poorly, succeeding in only 18% of trials, while the methods leveraging CNN annotations succeeded over 4 times as often.”
Quote in the paperSubstantial claim- Rests on an assumption: Human-assisted annotations of the intraoperative image are typically required to efficiently solve complex registration problems during navigation.
- Rests on an assumption: Manual initialization interferes with the surgical workflow and diminishes advantages gained from navigation.
May combine two claims.
InterpretationreasoningMethod 3 was robust to poor initializations, which most likely led to the larger number of successful pelvis registrations compared with method 2.
“Method 3 was robust to poor initializations, which most likely resulted in the larger number of successful pelvis registrations compared to those resulting from method 2.”
Quote in the paperSubstantial claim- Rests on an assumption: Large errors in pose estimates may occur when a registration is not initialized sufficiently close to the actual pose of an object.
Classified as probably not a substantial claim. May combine two claims.
Study design and conductown dataTest data sets consisted of images from a single specimen, and no images from that specimen were present in the training and validation data.
“Test data sets are comprised of images collected from a single specimen, and no images derived from this specimen are present in the training and validation data.”
Quote in the paperSubstantial claimMay combine two claims.
Interpretationown dataDespite a fairly large false-negative detection rate of 17%, an average of 7 landmarks per image were detected, allowing methods 2 and 3 to perform well.
“Despite the fairly large false-negative detection rate of 17%, an average of 7 landmarks per image were detected, allowing methods 2 and 3 to perform well.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Study design and conductnone givenManual inspection was performed to prune images corresponding to failed registrations from the training dataset.
“Manual inspection is performed so that images corresponding to failed registrations are pruned from the dataset.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim.
Study design and conductnone givenThe 2D labels and landmarks for each projection were manually inspected and verified.
“The 2D labels and landmarks for each projection are manually inspected and verified.”
Quote in the paperSubstantial claimMay combine two claims.
InterpretationreasoningThe pelvis's fluoroscopic appearance is more sensitive to rotational changes than the femur's, consistent with the smaller rotation errors observed for pelvis registrations.
“The pelvis’ fluoroscopic appearance is thus more sensitive to rotational changes than that of the femur, which is consistent with the smaller rotation errors observed with registrations of the pelvis and shown in Table 2.”
Quote in the paperNot a substantial claim- Rests on an assumption: The pelvis is a larger, more complex shape than the femur, occupying larger regions of the fluoroscopic views.
- Rests on an assumption: The pelvis's fluoroscopic appearance is more sensitive to rotational changes than that of the femur.
May combine two claims.
InterpretationreasoningThe influence of pelvis translation errors on these estimates was minimal because nearly all pelvis translation error was in the depth direction, which has minor impact on single-view appearance.
“The influence of pelvis translation errors on these estimates was minimal, since nearly all of the pelvis translation error was found in the depth direction, which has a minor impact on single-view appearance.”
Quote in the paperSubstantial claimMay combine two claims.
InterpretationreasoningThis prior knowledge is incorporated into the registration strategies and causes mostly small femoral translations to be reported, resulting in the small translation errors for femur registrations.
“This prior knowledge is incorporated into the registration strategies and causes mostly small femoral translations to be reported, which results in the small translation errors for femur registrations listed in Table 2.”
Quote in the paperSubstantial claim- Rests on an assumption: For intact hip anatomy, connective tissues joining each femoral head to the acetabular regions of the pelvis cause each femur and the pelvis to mostly translate together.
- Rests on an assumption: As a result, the relative pose of a femur with respect to the pelvis contains very little translation.
May combine two claims.
InterpretationreasoningThis robustness removes the need for manual initialization and allows navigation to be more naturally integrated into existing surgical workflows.
“This is a clinically relevant result, as this robustness removes the need for manual initialization and allows navigation to be more naturally integrated into existing surgical workflows.”
Quote in the paperSubstantial claim
The paper relies on this without support.
Human-assisted annotations of the intraoperative image are typically required to efficiently solve complex registration problems during navigation.
“In order to efficiently solve the complex registration problems presented during navigation, human-assisted annotations of the intraoperative image are typically required.”
The paper relies on this without support.
Manual initialization interferes with the surgical workflow and diminishes advantages gained from navigation.
“This manual initialization interferes with the surgical workflow and diminishes any advantages gained from navigation.”
The paper relies on this without support.
Manual initialization of processing may interrupt surgical workflows and negatively affect patient outcomes through increased operating time or blood loss.
“Although these systems report favorable navigation-related results, the manual initialization of processing may interrupt surgical workflows and negatively affect patient outcomes due to increased operating time or blood loss.”
The paper relies on this without support.
Large errors in pose estimates may occur when a registration is not initialized sufficiently close to the actual pose of an object.
“Large errors in pose estimates may occur when a registration is not initialized sufficiently close to the actual pose of an object.”
The paper relies on this without support.
Quality initializations are derived from manual human input, often through annotated landmark locations in the fluoroscopic image.
“Quality initializations are derived from some manual human input, often through annotated landmark locations in the fluoroscopic image.”
The paper relies on this without support.
For intact hip anatomy, connective tissues joining each femoral head to the acetabular regions of the pelvis cause each femur and the pelvis to mostly translate together.
“For the intact hip anatomy that is considered in this paper, the connective tissues joining each femoral head to the acetabular regions of the pelvis cause each femur and the pelvis to mostly translate together.”
The paper relies on this without support.
As a result, the relative pose of a femur with respect to the pelvis contains very little translation.
“As a result, the relative pose of a femur with respect to the pelvis contains very little translation.”
The paper relies on this without support.
The pelvis is a larger, more complex shape than the femur, occupying larger regions of the fluoroscopic views.
“In comparison to the femur, the pelvis is a larger, more complex shape, which occupies larger regions of the fluoroscopic views.”
The paper relies on this without support.
The pelvis's fluoroscopic appearance is more sensitive to rotational changes than that of the femur.
“The pelvis’ fluoroscopic appearance is thus more sensitive to rotational changes than that of the femur”
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Landmark regularization achieved an SMSR of 1.7%.
“Landmark regularization [5], an extension of xReg that uses predicted 2D landmarks as an additional loss term during optimization from the fixed initialization, achieved an SMSR of 1.7%.”
From Rapid patient-specific neural networks for intraoperative X-ray to volume registration (2025) · cites this paper as [5]
The Methods (Intraoperative Registration; page unavailable) describe using predicted 2D landmarks in a reprojection regularizer for intensity-based registration. Neither that passage nor the Results (Abstract; page unavailable) report an SMSR of 1.7%; the Results instead report registration within 1° for 86% of images, a different measure. (LLM, conf=0.86)
Experimental — accuracy not yet measuredIn the Femur and Ljubljana datasets, there are no 2D segmentation masks from which to regress annotated fiducials, so landmark localization is not feasible with the UNet model proposed in [5].
“For example, in the Femur and Ljubljana datasets, there are no 2D segmentation masks from which to regress annotated fiducials, so landmark localization is not feasible with the UNet model proposed in [5].”
From Rapid patient-specific neural networks for intraoperative X-ray to volume registration (2025) · cites this paper as [5]
The Methods sections on Network Architecture and Training and Extracting Landmark Locations describe a U-Net with segmentation and landmark-heatmap paths that uses both outputs to locate landmarks. The paper does not discuss the Femur or Ljubljana datasets, establish whether they lack 2D masks, or conclude that landmark localization on them is infeasible. Page numbers are unavailable in our copy. (LLM, conf=0.86)
Experimental — accuracy not yet measuredNeural networks trained on small datasets for 2D/3D registration fail on new patients or require impractical landmark supervision.
“neural networks trained on small datasets fail on new patients or require impractical landmark supervision”
From Intraoperative 2D/3D Image Registration via Differentiable X-Ray Rendering (2024) · cites this paper as [25]
The Discussion and Conclusion (page unavailable) reports poor registration on one held-out specimen whose C-arm poses differed from those in training, supporting a limited generalization concern—not a general claim that networks trained on small datasets fail on new patients. The Methods, Training Dataset Creation (page unavailable) describes automatically generating landmark labels and manually verifying them; it does not establish that landmark supervision is impractical. (LLM, conf=0.84)
Experimental — accuracy not yet measuredLandmark-based localization requires expert knowledge of landmarks visible on X-ray and manual annotation of 3D landmarks for every new CT volume.
“Such methods require expert knowledge of landmarks that are visible on X-ray for the specific surgical application and manual annotation of 3D landmarks for every new CT volume [25].”
From Intraoperative 2D/3D Image Registration via Differentiable X-Ray Rendering (2024) · cites this paper as [25]
Methods > Data Preprocessing reports that 14 anatomical landmarks were manually annotated in 3D, and Experiments and Results > Data Collection and Training Dataset Creation says landmarks were manually digitized for each CT scan in the study. These passages support the manual-annotation part, but do not establish that landmark-based methods generally require expert knowledge of X-ray-visible landmarks for each surgical application or manual annotation for every new CT volume. (LLM, conf=0.83)
Experimental — accuracy not yet measuredThe CMAES benchmark method for pose estimation of bone anatomy and surgical devices has been tested to meet clinical requirements in various orthopedic applications.
“Due to its well-known robustness to local minima, the CMAES benchmark method for pose estimation of bone anatomy and surgical devices has been tested to meet clinical requirements in various orthopedic applications, including osteotomy [2], [13], femoroplasty [41], [4], core decompression of the hip [42], and transforaminal lumbar epidural injections [43], etc.”
From A Fully Differentiable Framework for 2D/3D Registration and the Projective Spatial Transformers (2024) · cites this paper as [13]
The Methods (Training Dataset Creation and Intraoperative Registration; pages unavailable) use CMA-ES for pelvis and femur registration, and the Results evaluate registration on cadaveric hip fluoroscopy. The paper does not test CMA-ES for surgical-device pose estimation or show that it meets clinical requirements across the listed orthopedic applications; the Discussion describes registration of instruments as future research. (LLM, conf=0.88)
Experimental — accuracy not yet measured
The average runtime of a conventional 2D/3D registration algorithm using landmark-based initialization is 7 seconds.
“As a comparison, the average runtime of a conventional 2D/3D registration algorithm using landmark-based initialization is 7 seconds [13].”
From A Fully Differentiable Framework for 2D/3D Registration and the Projective Spatial Transformers (2024) · cites this paper as [13]
The Intraoperative Registration results report a mean runtime of 7.2 ± 0.7 seconds for method 3, so the seven-second figure is a reasonable rounding. But the Methods describe method 3 as a registration approach incorporating detected landmarks into a robust reprojection regularizer, not simply a conventional algorithm using landmark-based initialization; the paper does not assign that runtime to its naive method 1. (LLM, conf=0.86)
Experimental — accuracy not yet measuredFeature-based 2D/3D registration methods need to define meaningful landmark features manually and do not learn registration in an end-to-end fashion.
“However, these methods need to define meaningful landmark features manually and do not learn registration in an end-to-end fashion.”
From A Fully Differentiable Framework for 2D/3D Registration and the Projective Spatial Transformers (2024) · cites this paper as [13]
Methods > Data Preprocessing states that 14 anatomical landmarks were manually annotated in 3D. Methods > Intraoperative Registration describes a separate optimization step rather than end-to-end learned registration. But the paper’s proposed approach uses a CNN to locate landmarks in fluoroscopy and requires no human assistance intraoperatively, so it does not support the unqualified claim that feature-based methods need manual landmark definition during registration. The supplied copy gives no page numbers. (LLM, conf=0.82)
Experimental — accuracy not yet measuredxReg did not achieve sufficient accuracy, producing SMSRs of just 0.3% and 3.3% for the fixed and landmark initializations, respectively.
“xReg did not achieve sufficient accuracy, producing SMSRs of just 0.3% and 3.3% for the fixed [20] and landmark initializations [5], respectively (Fig. 4a).”
From Rapid patient-specific neural networks for intraoperative X-ray to volume registration (2025) · cites this paper as [5]
The full pairwise analysis yields a ranking of patient-specific methods (rank #1) > foundation model (rank #3) > landmark initialization (rank #4) > landmark regularization (rank #5) > fixed initialization (rank #6), with all differences significant after correction.
“The full pairwise analysis yields a clear ranking: patient-specific methods (rank #1) > foundation model (rank #3) > landmark initialization [6] (rank #4) > landmark regularization [6] (rank #5) > fixed initialization [4,5] (rank #6), with all differences significant after correction.”
From Rapid patient-specific neural networks for intraoperative X-ray to volume registration (2025) · cites this paper as [5]
Recent approaches train neural networks to segment intraoperative X-rays and localize predefined landmarks from segmentation masks.
“Recent approaches train neural networks to segment intraoperative X-rays and localize predefined landmarks from the segmentation masks [17, 25, 27, 37, 39].”
From Intraoperative 2D/3D Image Registration via Differentiable X-Ray Rendering (2024) · cites this paper as [25]
A PnP solver can be subject to poor initialization especially if less than four landmarks are visible in the 2D image.
“These landmarks are used with a PnP solver to predict the camera pose, which can be subject to poor initialization especially if less than four landmarks are visible in the 2D image [25].”
From Intraoperative 2D/3D Image Registration via Differentiable X-Ray Rendering (2024) · cites this paper as [25]
Groundtruth poses of the pelvis were estimated using the comprehensive image intensity-based 2D/3D registration pipeline described in [13].
“Groundtruth poses of the pelvis were estimated using the comprehensive image intensity-based 2D/3D registration pipeline described in [13].”
From A Fully Differentiable Framework for 2D/3D Registration and the Projective Spatial Transformers (2024) · cites this paper as [13]
There are popular feature-based 2D/3D registration methods, such as solving a PnP problem using corresponding anatomical landmarks, which may be complementary to intensity-based registration methods.
“We are aware that there are popular feature-based 2D/3D registration methods, such as solving a PnP problem using corresponding anatomical landmarks [13], [16], [17], which may be complementary to the intensity-based registration methods.”
From A Fully Differentiable Framework for 2D/3D Registration and the Projective Spatial Transformers (2024) · cites this paper as [13]
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474 · see citing works
U-Net: Convolutional Networks for Biomedical Image Segmentation
Ronneberger, O., Fischer, P., & Brox, T. (2015) · Lecture notes in computer science · doi:10.1007/978-3-319-24574-4_28 · Cited by 99,404 · cited 2× · see citing works
“For the encoder–decoder in this paper, we adopt a 6-level U-Net [27] design with 32 features at the top level and 1024 features at the bottom.”
MethodFully convolutional networks for semantic segmentationPreprint
Long, J., Shelhamer, E., & Darrell, T. (2014) · Computer Vision and Pattern Recognition · doi:10.48550/arxiv.1605.06211 · Cited by 43,601 · see citing works
Fully convolutional networks for semantic segmentation
Long, J., Shelhamer, E., & Darrell, T. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298965 · Cited by 38,834 · see citing works
Storn, R., & Price, K. (1997) · Journal of Global Optimization · doi:10.1023/a:1008202821328 · Cited by 29,569 · cited 1× · see citing works
“An attempt is first made to register the pelvis using a mixture of the Differential Evolution [23], exhaustive grid search, Particle Swarm [24], Covariance Matrix Adaptation: Evolutionary Search (CMA-ES) [25], and Bounded Optimization by Quadratic Approximation (BOBYQA) [26] optimization strategies at multiple resolut…”
MethodBatch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
Ioffe, S., & Szegedy, C. (2015) · arXiv · doi:10.48550/arxiv.1502.03167 · Cited by 27,818 · see citing works
Multiple View Geometry in Computer Vision
Hartley, R., & Zisserman, A. (2004) · doi:10.1017/cbo9780511811685 · Cited by 21,070 · see citing works
V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation
Milletari, F., Navab, N., & Ahmadi, S.-A. (2016) · 2016 Fourth International Conference on 3D Vision (3DV) · doi:10.1109/3dv.2016.79 · Cited by 11,931 · cited 1× · see citing works
“The differentiable dice score [28] is computed for each class and then averaged.”
MethodFully convolutional networks for semantic segmentation
Shelhamer, E., Long, J., & Darrell, T. (2017) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2016.2572683 · Cited by 11,898 · cited 1× · see citing works
“Convolutional neural networks (CNNs) have excelled at detecting landmarks and performing semantic segmentation when sufficiently large annotated datasets are available for training [10, 11].”
BackgroundA modified particle swarm optimizer
Shi, Y., & Eberhart, R. (1998) · 1998 IEEE International Conference on Evolutionary Computation Proceedings. IEEE World Congress on Computational Intelligence (Cat. No.98TH8360) · doi:10.1109/icec.1998.699146 · Cited by 11,497 · cited 1× · see citing works
“An attempt is first made to register the pelvis using a mixture of the Differential Evolution [23], exhaustive grid search, Particle Swarm [24], Covariance Matrix Adaptation: Evolutionary Search (CMA-ES) [25], and Bounded Optimization by Quadratic Approximation (BOBYQA) [26] optimization strategies at multiple resolut…”
MethodStacked Hourglass Networks for Human Pose Estimation
Newell, A., Yang, K., & Deng, J. (2016) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-46484-8_29 · Cited by 5,634 · cited 1× · see citing works
“Convolutional neural networks (CNNs) have excelled at detecting landmarks and performing semantic segmentation when sufficiently large annotated datasets are available for training [10,11].”
BackgroundCompletely Derandomized Self-Adaptation in Evolution Strategies
Hansen, N., & Ostermeier, A. (2001) · Evolutionary Computation · doi:10.1162/106365601750190398 · Cited by 4,397 · cited 1× · see citing works
“An attempt is first made to register the pelvis using a mixture of the Differential Evolution [23], exhaustive grid search, Particle Swarm [24], Covariance Matrix Adaptation: Evolutionary Search (CMA-ES) [25], and Bounded Optimization by Quadratic Approximation (BOBYQA) [26] optimization strategies at multiple resolut…”
MethodThe BOBYQA algorithm for bound constrained optimization without derivatives
Powell, M. J. D. (2009) · Cited by 1,094 · see citing works
A Review of 3d/2d Registration Methods for Image-Guided Interventions
Markelj, P., Tomaževič, D., Likar, B., & Pernuš, F. (2012) · Medical Image Analysis · doi:10.1016/j.media.2010.03.005 · Cited by 746 · cited 1× · see citing works
“In order to report object poses accurately, fluoroscopic navigation systems rely on 2D/3D registrations between intraoperative 2D images and the appropriate 3D models [9].”
A CNN Regression Approach for Real-Time 2D/3D Registration
Miao, S., Wang, Z. J., & Liao, R. (2016) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2016.2521800 · Cited by 514 · see citing works
Concurrent Segmentation and Localization for Tracking of Surgical Instruments
Laina, I., Rieke, N., Rupprecht, C., Vizcaíno, J. P., Eslami, A., Tombari, F., & Navab, N. (2017) · Lecture notes in computer science · doi:10.1007/978-3-319-66185-8_75 · Cited by 165 · cited 3× · see citing works
“By reusing segmentation features from an encoder–decoder style network for the computation of landmark heatmaps, Laina was able to automatically annotate segmentation labels and landmark locations of tools used in laparoscopy and retinal microsurgery [17].”
MethodFully Automatic and Real-Time Catheter Segmentation in X-Ray Fluoroscopy
Ambrosini, P., Ruijters, D., Niessen, W. J., Moelker, A., & van Walsum, T. (2017) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-66185-8_65 · Cited by 102 · see citing works
Slotkin, E. M., Patel, P. D., & Suárez, J. C. (2015) · The Journal of Arthroplasty · doi:10.1016/j.arth.2015.03.046 · Cited by 85 · cited 1× · see citing works
“and subject to an extensive learning curve [2,3].”
BackgroundEnabling machine learning in X-ray-based procedures via realistic simulation of image formation
Unberath, M., Zaech, J.-N., Gao, C., Bier, B., Goldmann, F., Lee, S. C., Fotouhi, J., Taylor, R., Armand, M., & Navab, N. (2019) · International Journal of Computer Assisted Radiology and Surgery · doi:10.1007/s11548-019-02011-2 · Cited by 63 · cited 1× · see citing works
“By augmenting actual fluoroscopy with realistic synthetic fluoroscopy [30] during training, we believe that quality performance at these “missing” poses may be achieved.”
BackgroundSelf-citationReyneke, C., Lüthi, M., Burdin, V., Douglas, T. S., Vetter, T. R., & Mutsvangwa, T. (2018) · IEEE Reviews in Biomedical Engineering · doi:10.1109/rbme.2018.2876450 · Cited by 61 · cited 1× · see citing works
“The automated annotation and registration techniques proposed in this paper could streamline intraoperativeworkflows related to intact hip anatomy, such as osteotomy planning [8], robotic drilling [31], and 3D reconstructions of bone [32] or implanted tracking fiducials [33].”
MethodRole of Navigation in Total Hip Arthroplasty
Kelley, T. C., & Swank, M. L. (2009) · Journal of Bone and Joint Surgery · doi:10.2106/jbjs.h.01463 · Cited by 57 · cited 1× · see citing works
“Systems leveraging fluoroscopy have been developed for total hip arthroplasty [4], hip resurfacing [5], cement injection [6], and osteotomies of the acetabulum [7] or proximal femur [8].”
BackgroundLearning to detect anatomical landmarks of the pelvis in X-rays from arbitrary views
Bier, B., Goldmann, F., Zaech, J.-N., Fotouhi, J., Hegeman, R., Grupp, R., Armand, M., Osgood, G., Navab, N., Maier, A., et al. (2019) · International Journal of Computer Assisted Radiology and Surgery · doi:10.1007/s11548-019-01975-5 · Cited by 43 · cited 2× · see citing works
“Using a large collection of simulated fluoroscopy, Bier trained CNNs to annotate anatomical landmarks of the pelvis [19].”
MethodSelf-citationWoerner, M. L., Sendtner, E., Springorum, R., et al. (2016) · Acta Orthopaedica · doi:10.3109/17453674.2015.1137182 · Cited by 42 · cited 1× · see citing works
“Minimally invasive surgical interventions of the hip manipulate, modify, or augment anatomical structures which are hidden or not reliably visible [1].”
BackgroundIntraoperative stent segmentation in X-ray fluoroscopy for endovascular aortic repair
Breininger, K., Albarqouni, S., Kurzendorfer, T., Pfister, M., Kowarschik, M., & Maier, A. (2018) · International Journal of Computer Assisted Radiology and Surgery · doi:10.1007/s11548-018-1779-6 · Cited by 38 · cited 1× · see citing works
“However, since existing large-scale hip datasets have focused on 3D image modalities and pre and postoperative radiography, rather than intraoperative fluoroscopy [12], applications of CNNs to fluoroscopy have been mostly limited to recognizing surgical instruments and tools [13–16].”
BackgroundPose Estimation of Periacetabular Osteotomy Fragments With Intraoperative X-Ray Navigation
Grupp, R. B., Hegeman, R. A., Murphy, R. J., Alexander, C. P., Otake, Y., McArthur, B. A., Armand, M., & Taylor, R. H. (2020) · IEEE Transactions on Biomedical Engineering · doi:10.1109/tbme.2019.2915165 · Cited by 38 · cited 5× · see citing works
“These landmarks were previously identified as being useful for obtaining initial registration estimates of the pelvis [7].”
BackgroundMethodInfluentialSelf-citationTowards Fully Automatic X-Ray to CT Registration
Esteban, J., Grimm, M., Unberath, M., Zahnd, G., & Navab, N. (2019) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-32226-7_70 · Cited by 38 · cited 3× · see citing works
“No analysis on actual fluoroscopy was conducted in [20].”
BackgroundSelf-citationFiducial-Free 2D/3D Registration for Robot-Assisted Femoroplasty
Gao, C., Farvardin, A., Grupp, R. B., Bakhtiarinejad, M., Ma, L., Thies, M., Unberath, M., Taylor, R. H., & Armand, M. (2020) · IEEE Transactions on Medical Robotics and Bionics · doi:10.1109/tmrb.2020.3012460 · Cited by 36 · see citing works
Self-citationFluoroscopic navigation system for hip surface replacement
Belei, P., Skwara, A., de la Fuente, M., et al. (2007) · Computer Aided Surgery · doi:10.3109/10929080701336207 · Cited by 27 · cited 1× · see citing works
“Systems leveraging fluoroscopy have been developed for total hip arthroplasty [4], hip resurfacing [5], cement injection [6], and osteotomies of the acetabulum [7] or proximal femur [8].”
BackgroundSurgical advances in periacetabular osteotomy for treatment of hip dysplasia in adults
Troelsen, A. (2009) · Acta Orthopaedica · doi:10.1080/17453690610046585 · Cited by 25 · cited 1× · see citing works
“and subject to an extensive learning curve [2,3].”
BackgroundNikou, C., Jaramaz, B., DiGioia, A. M., et al. (2000) · Lecture notes in computer science · doi:10.1007/978-3-540-40899-4_127 · Cited by 25 · cited 1× · see citing works
“SPS landmarks [21] and is later used to estimate nominal anterior/posterior (AP) poses and as a reference coordinate frame during registration.”
MethodGrupp, R. B., Armand, M., & Taylor, R. H. (2018) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-01201-4_17 · Cited by 23 · cited 1× · see citing works
“In this paper, S is defined as the weighted sum of normalized cross-correlations of 2D image gradients computed over image patches [22].”
MethodSelf-citationMulti-task Localization and Segmentation for X-Ray Guided Planning in Knee Surgery
Kordon, F., Fischer, P., Privalov, M., et al. (2019) · Lecture notes in computer science · doi:10.1007/978-3-030-32226-7_69 · Cited by 20 · cited 1× · see citing works
“Kordon demonstrated that a CNN, trained from 149 manually annotated preoperative radiographs, could successfully segment the four bones of the knee joint and locate two anatomical landmarks and a surgically relevant line [18].”
BackgroundIntraoperative, fluoroscopy-based planning for complex osteotomies of the proximal femur
Gottschling, H., Roth, M., Schweikard, A., et al. (2005) · International Journal of Medical Robotics and Computer Assisted Surgery · doi:10.1002/rcs.29 · Cited by 19 · cited 2× · see citing works
“The automated annotation and registration techniques proposed in this paper could streamline intraoperative workflows related to intact hip anatomy, such as osteotomy planning [8], robotic drilling [31], and 3D reconstructions of bone [32] or implanted tracking fiducials [33].”
BackgroundMethodLocalizing dexterous surgical tools in X-ray for image-based navigation
Gao, C., Unberath, M., Taylor, R., & Armand, M. (2019) · arXiv · doi:10.48550/arxiv.1901.06672 · Cited by 15 · cited 4× · see citing works
“Gao also leveraged this approach for the localization of a dexterous continuum manipulator in fluoroscopy [16].”
MethodInfluentialSelf-citationFiducial-free 2D/3D registration of the proximal femur for robot-assisted femoroplasty
Gao, C., Grupp, R. B., Unberath, M., Taylor, R. H., & Armand, M. (2020) · Medical Imaging 2020: Image-Guided Procedures, Robotic Interventions, and Modeling · doi:10.1117/12.2550992 · Cited by 14 · cited 1× · see citing works
“The automated annotation and registration techniques proposed in this paper could streamline intraoperativeworkflows related to intact hip anatomy, such as osteotomy planning [8], robotic drilling [31], and 3D reconstructions of bone [32] or implanted tracking fiducials [33].”
MethodSelf-citationFluoroscopic navigation system for hip surface replacement
Belei, P., Skwara, A., de la Fuente, M., et al. (2007) · Computer Aided Surgery · doi:10.1080/10929080701336207 · Cited by 7 · see citing works
Malan, D. F., van der Walt, S., Raidou, R. G., et al. (2015) · International Journal of Computer Assisted Radiology and Surgery · doi:10.1007/s11548-015-1252-8 · Cited by 5 · cited 1× · see citing works
“Systems leveraging fluoroscopy have been developed for total hip arthroplasty [4], hip resurfacing [5], cement injection [6], and osteotomies of the acetabulum [7] or proximal femur [8].”
BackgroundIntraoperative, fluoroscopy-based planning for complex osteotomies of the proximal femur
Gottschling, H., Roth, M., Schweikard, A., et al. (2005) · International Journal of Medical Robotics and Computer Assisted Surgery · doi:10.1581/mrcas.2005.010303 · Cited by 2 · see citing works
Construction and Application of Large-Scale Image Database in Orthopedic Surgery
Otake, Y., Takao, M., Yokota, F., Fukuda, N., Uemura, K., Sugano, N., & Sato, Y. (2018) · Computer Assisted Orthopaedic Surgery for Hip and Knee · doi:10.1007/978-981-10-5245-3_16 · Cited by 1 · cited 1× · see citing works
“However, since existing large-scale hip datasets have focused on 3D image modalities and pre and postoperative radiography, rather than intraoperative fluoroscopy [12], applications of CNNs to fluoroscopy have been mostly limited to recognizing surgical instruments and tools [13–16].”
BackgroundSelf-citationGrupp, R., Murphy, R., Hegeman, R., et al. (2019) · arXiv · doi:10.48550/arxiv.1910.10187
Self-citation
Flagged first, then by citations
4 of the 44 references couldn’t be matched to a known paper.
Cited by
Sorted by citations
Gao, C., Killeen, B. D., Hu, Y., Grupp, R. B., Taylor, R. H., Armand, M., & Unberath, M. (2023) · Nature Machine Intelligence · doi:10.1038/s42256-023-00629-1 · Cited by 123 · see citing works
“This accuracy was tested to be effective in initializing the 2D/3D pelvis registration and achieving less than 1° error for 86% of the images 12 .”
BackgroundMethodInfluentialSelf-citationColonoscopy 3D video dataset with paired depth from 2D-3D registration
Bobrow, T. L., Golhar, M., Vijayan, R. C., et al. (2023) · Medical Image Analysis · doi:10.1016/j.media.2023.102956 · Cited by 113 · see citing works
Unberath, M., Gao, C., Hu, Y., Judish, M., Taylor, R. H., Armand, M., & Grupp, R. (2021) · Frontiers in Robotics and AI · doi:10.3389/frobt.2021.716007 · Cited by 102 · see citing works
“…to define correspondences with the respective 3D locations, which allows for either, explicit determination of ˆ y v = { K v , v T x } using PnP [41, 45] or stereo-reconstruction following 3D-3D matching [38, 44], or for the introduction of soft re-projection constraints as a regularizing term [35].”
BackgroundMethodInfluentialSelf-citationEvolution of Surgical Robot Systems Enhanced by Artificial Intelligence: A Review
Liu, Y., Wu, X., Sang, Y., Zhao, C., Wang, Y., Shi, B., & Fan, Y. (2024) · Advanced Intelligent Systems · doi:10.1002/aisy.202300268 · Cited by 105 · see citing works
Intraoperative 2D/3D Image Registration via Differentiable X-Ray Rendering
Gopalakrishnan, V., Dey, N., & Golland, P. (2024) · 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52733.2024.01108 · Cited by 48 · see citing works
Generalizing Spatial Transformers to Projective Geometry with Applications to 2D/3D Registration
Gao, C., Liu, X., Gu, W., Killeen, B., Armand, M., Taylor, R., & Unberath, M. (2020) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-59716-0_32 · Cited by 46 · see citing works
“Following [2], we formulate an intensity-based 2D/3D registration problem with a pre-operative CT volume V , Digitally Reconstructed Radiograph (DRR) projection operator P , pose parameter θ, a fixed target image If , and a similarity metric loss LS :”
MethodSelf-citationA Fully Differentiable Framework for 2D/3D Registration and the Projective Spatial Transformers
Gao, C., Feng, A., Liu, X., Taylor, R. H., Armand, M., & Unberath, M. (2024) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2023.3299588 · Cited by 36 · see citing works
“Our real hip X-ray data were selected from the cadaveric X-rays released by Grupp et al. [13].”
BackgroundMethodInfluentialSelf-citationFiducial-Free 2D/3D Registration for Robot-Assisted Femoroplasty
Gao, C., Farvardin, A., Grupp, R. B., Bakhtiarinejad, M., Ma, L., Thies, M., Unberath, M., Taylor, R. H., & Armand, M. (2020) · IEEE Transactions on Medical Robotics and Bionics · doi:10.1109/tmrb.2020.3012460 · Cited by 36 · see citing works
“during fluoroscopic navigation of the hip [25].”
Self-citationPose-Dependent Weights and Domain Randomization for Fully Automatic X-Ray to CT Registration
Grimm, M., Esteban, J., Unberath, M., & Navab, N. (2021) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2021.3073815 · Cited by 34 · see citing works
“The scenario in the present paper is similar to the scenario of [6].”
InfluentialSelf-citationSelf-Supervised 2D/3D Registration for X-Ray to CT Image Fusion
Jaganathan, S., Kukla, M., Wang, J., Shetty, K., & Maier, A. (2023) · 2023 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) · doi:10.1109/wacv56688.2023.00281 · Cited by 30 · see citing works
Fluoroscopic Navigation for a Surgical Robotic System Including a Continuum Manipulator
Gao, C., Phalen, H., Sefati, S., Ma, J. H., Taylor, R. H., Unberath, M., & Armand, M. (2021) · IEEE Transactions on Biomedical Engineering · doi:10.1109/tbme.2021.3097631 · Cited by 31 · see citing works
Self-citationFluoroscopy-Guided Robotic System for Transforaminal Lumbar Epidural Injections
Gao, C., Phalen, H., Margalit, A., Ma, J. H., Ku, P., Unberath, M., Taylor, R. H., Jain, A., & Armand, M. (2022) · IEEE Transactions on Medical Robotics and Bionics · doi:10.1109/tmrb.2022.3196321 · Cited by 25 · see citing works
“Different clinical applications have different algorithmic designs for fiducial-free navigation, such as automatic pelvic landmark detection [21], multi-modality registration [22], etc.”
BackgroundSelf-citationA Patient-Specific Self-supervised Model for Automatic X-Ray/CT Registration
Zhang, B., Faghihroohi, S., Azampour, M. F., Liu, S., Ghotbi, R., Schunkert, H., & Navab, N. (2023) · Lecture Notes in Computer Science · doi:10.1007/978-3-031-43996-4_49 · Cited by 27 · see citing works
Take a shot! Natural language control of intelligent robotic X-ray systems in surgery
Killeen, B. D., Chaudhary, S., Osgood, G. M., & Unberath, M. (2024) · International Journal of Computer Assisted Radiology and Surgery · doi:10.1007/s11548-024-03120-3 · Cited by 19 · see citing works
“These anatomical landmarks are well defined in the literature [20, 22, 25, 28], consisting of identifiable points on the surface of the pelvis, such as bony protrusions.”
Self-citationGeng, H., Xiao, D., Yang, S., Fan, J., Fu, T., Lin, Y., Bai, Y., Ai, D., Song, H., Wang, Y., et al. (2023) · Physics in Medicine and Biology · doi:10.1088/1361-6560/acede5 · Cited by 17 · see citing works
Extended Capture Range of Rigid 2D/3D Registration by Estimating Riemannian Pose Gradients
Gu, W., Gao, C., Grupp, R., Fotouhi, J., & Unberath, M. (2020) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-59861-7_29 · Cited by 16 · see citing works
“While pre-processing or style-transfer may be feasible, unsupervised domain adaptation or retraining on small clinical datasets that could be annotated using approach similar to [9] can also be considered.”
BackgroundSelf-citationIn silico simulation: a key enabling technology for next-generation intelligent surgical systems
Killeen, B. D., Cho, S. M., Armand, M., Taylor, R. H., & Unberath, M. (2023) · Progress in Biomedical Engineering · doi:10.1088/2516-1091/acd28b · Cited by 16 · see citing works
“Although specific spatial annotations, such as camera pose, tool-to-tissue relationships, and anatomical shape, may be obtained from real images through post hoc analysis [36], by incorporating robot kinematic models [37, 38], or by leveraging the shape of known components [39, 40], the assumptions made by these appro…”
Self-citationZhou, C., Cha, T., Peng, Y. F., & Li, G. (2021) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2021.104923 · Cited by 11 · see citing works
Segment any medical model extended
Liu, Y., Zhang, J., Diaz‐Pinto, A., et al. (2024) · SPIE Medical Imaging: Image Processing · doi:10.1117/12.3001069 · Cited by 11 · see citing works
“Example applications include tumor diagnosis, 1 dental model reconstruction, 2 jaw transplantation, 3 brain mapping, 4 lesion detection, 5 bone abnormality identification, 6 vascular structure analysis, 7 2D/3D registration, 8,9 and fluoroscopy-guided procedures.”
Self-citationFully automatic tracking of native glenohumeral kinematics from stereo-radiography
Burton, W. S., Crespo, I. R., Andreassen, T. E., Pryhoda, M. K., Jensen, A., Myers, C. A., Shelburne, K. B., Banks, S. A., & Rullkoetter, P. J. (2023) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2023.107189 · Cited by 10 · see citing works
Killeen, B. D., Gao, C., Oguine, K. J., et al. (2023) · International Journal of Computer Assisted Radiology and Surgery · doi:10.1007/s11548-023-02941-y · Cited by 11 · see citing works
“[14] demonstrates the sim-to-real performance of the DNN, but because these images are acquired from views not planned by our robotic imaging technique, they do not capture the advantage of viewpoint planning as in our system.”
BackgroundMethodInfluentialSelf-citationC-Arm Positioning for Spinal Standard Projections in Different Intra-operative Settings
Kausch, L., Thomas, S., Kunze, H., Norajitra, T., Klein, A., Barbari, J. S. E., Privalov, M., Vetter, S. Y., Mahnken, A. H., Maier‐Hein, L., et al. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-87202-1_34 · Cited by 11 · see citing works
Zhang, J., Mazurowski, M. A., Allen, B. C., & Wildman‐Tobriner, B. (2023) · Artificial Intelligence in Medicine · doi:10.1016/j.artmed.2023.102553 · Cited by 10 · see citing works
Visualization in 2D/3D registration matters for assuring technology-assisted image-guided surgery
Cho, S. M., Grupp, R. B., Gómez, C., et al. (2023) · International Journal of Computer Assisted Radiology and Surgery · doi:10.1007/s11548-023-02888-0 · Cited by 10 · see citing works
“In spine and orthopedic minimally invasive operations, such as total hip arthroplasty or osteotomies of the pelvis or femur, 2D/3D registration has been applied to help orient the current position of instruments relative to the planned trajectory, nearby vulnerable structures, and the ultimate target [6].”
BackgroundMethodInfluentialSelf-citationX-Ray to CT Rigid Registration Using Scene Coordinate Regression
Shrestha, P., Xie, C., Shishido, H., Yoshii, Y., & Kitahara, I. (2023) · Lecture Notes in Computer Science · doi:10.1007/978-3-031-43999-5_74 · Cited by 12 · see citing works
“Landmark estimation methods aim to solve for pose using correspondences between 3D landmark annotations and its estimated 2D projection points [7,3,5], while methods based on pose regression estimate the global camera pose in a single inference [12].”
MethodC-arm positioning for standard projections during spinal implant placement
Kausch, L., Thomas, S., Kunze, H., Norajitra, T., Klein, A., Ayala, L. A., Barbari, J. S. E., Mandelka, E., Privalov, M., Vetter, S. Y., et al. (2022) · Medical Image Analysis · doi:10.1016/j.media.2022.102557 · Cited by 9 · see citing works
Burton, W. S., Myers, C. A., Stefanovic, M., Shelburne, K. B., & Rullkoetter, P. J. (2024) · Journal of Biomechanics · doi:10.1016/j.jbiomech.2024.112066 · Cited by 9 · see citing works
Chênes, C., & Schmid, J. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-87231-1_44 · Cited by 9 · see citing works
Zhang, W., Zhao, L., Gou, H., Gong, Y., Zhou, Y., & Feng, Q. (2024) · Medical Image Analysis · doi:10.1016/j.media.2024.103283 · Cited by 9 · see citing works
Mixed reality interfaces for achieving desired views with robotic X-ray systems
Killeen, B. D., Winter, J., Gu, W., et al. (2022) · Computer Methods in Biomechanics and Biomedical Engineering Imaging & Visualization · doi:10.1080/21681163.2022.2154272 · Cited by 6 · see citing works
Self-citationA new 2D‐3D registration gold‐standard dataset for the hip joint based on uncertainty modeling
D’Isidoro, F., Chênes, C., Ferguson, S. J., & Schmid, J. (2021) · Medical Physics · doi:10.1002/mp.15124 · Cited by 6 · see citing works
Deep Learning Compatible Differentiable X-ray Projections for Inverse Rendering
Shetty, K., Birkhold, A., Strobel, N. K., et al. (2021) · Informatik aktuell · doi:10.1007/978-3-658-33198-6_70 · Cited by 6 · see citing works
“For evaluation, we used the dataset provided by Grupp et al. made up of the hip anatomy from 6 patients [7].”
MethodIntelligent control of robotic X-ray devices using a language-promptable digital twin
Killeen, B. D., Suresh, A., Gómez, C., Íñigo, B., Bailey, C., & Unberath, M. (2025) · International Journal of Computer Assisted Radiology and Surgery · doi:10.1007/s11548-025-03351-y · Cited by 5 · see citing works
Self-citationSun, W., Zhao, Y., Liu, J., et al. (2023) · International Journal of Computer Assisted Radiology and Surgery · doi:10.1007/s11548-023-02877-3 · Cited by 5 · see citing works
Mizoe, S., Otake, Y., Miyamoto, T., Soufi, M., Nakao, S., Tanaka, Y., & Sato, Y. (2021) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-87202-1_18 · Cited by 5 · see citing works
Self-citationAutomatic multi-view X-ray/CT registration using bone substructure contours
Flepp, R., Nissen, L., Sigrist, B., Nieuwland, A., Cavalcanti, N., Fürnstahl, P., Dreher, T., & Calvet, L. (2025) · International Journal of Computer Assisted Radiology and Surgery · doi:10.1007/s11548-025-03391-4 · Cited by 5 · see citing works
“As highlighted in prior research [10, 27, 28], single-view registration is significantly less reliable than multi-view approaches, particularly for symmetric objects like the femur.”
RayEmb: Arbitrary Landmark Detection in X-Ray Images Using Ray Embedding Subspace
Shrestha, P., Xie, C., Yoshii, Y., & Kitahara, I. (2024) · Lecture notes in computer science · doi:10.1007/978-981-96-0901-7_15 · Cited by 4 · see citing works
Multi-View 2D/3D Image Registration via Differentiable X-ray Rendering
Cui, Y., Song, R., Li, Y., Meng, M. Q. ‐., & Min, Z. (2024) · Procedia Computer Science · doi:10.1016/j.procs.2024.11.041 · Cited by 5 · see citing works
Towards visualising early-stage osteonecrosis using intraoperative imaging modalities
Liu, M., Gomez, A. M., Oni, J. K., Mears, S. C., & Armand, M. (2022) · Computer Methods in Biomechanics and Biomedical Engineering Imaging & Visualization · doi:10.1080/21681163.2022.2157329 · Cited by 5 · see citing works
“The generation of these images has proven valuable in performing 2D/3D registration of femurs/pelvis (Esteban et al. 2019; Grupp et al. 2020).”
BackgroundMethodSelf-citationGrupp, R. B., Murphy, R. J., Hegeman, R. A., et al. (2020) · Physics in Medicine and Biology · doi:10.1088/1361-6560/aba089 · Cited by 4 · see citing works
Self-citationCui, Y., Song, R., Li, Y., Meng, M. Q. ‐., & Min, Z. (2025) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra55743.2025.11128375 · Cited by 3 · see citing works
“In landmark-based registration, feature extractors establish the mapping relationship between 3D anatomical landmarks and 2D image landmarks and a Perspective-n-Point (PnP) solver [15] estimates the 2D image’s pose relative to the 3D structure [16], [17].”
Parametric Bi-invariant Learning for Improved Precision in 2D/3D Image Registration
Meng, R., Chen, C., Lu, M., Fu, X., Xiao, Y., Wang, K., Zou, Y., & Li, Y. (2025) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2025.107603 · Cited by 3 · see citing works
Autonomous Robotic Systems in Otolaryngology-Head and Neck Surgery
Sriram, S., Creighton, F. X., & Galaiya, D. J. (2024) · Otolaryngologic Clinics of North America · doi:10.1016/j.otc.2024.05.004 · Cited by 4 · see citing works
Burton, W. S., Myers, C. A., Stefanovic, M., Shelburne, K. B., & Rullkoetter, P. J. (2024) · Annals of Biomedical Engineering · doi:10.1007/s10439-024-03473-5 · Cited by 3 · see citing works
Generalizing Spatial Transformers to Projective Geometry with Applications to 2D/3D RegistrationPreprint
Gao, C., Liu, X., Gu, W., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2003.10987 · Cited by 3 · see citing works
“Following [2], we formulate an intensity-based 2D/3D registration problem with a pre-operative CT volume V , Digitally Reconstructed Radiograph (DRR) projection operator P , pose parameter θ, a fixed target image If , and a similarity metric loss LS :”
MethodSelf-citationJiang, Y., Li, P., Zhang, Y., Pei, Y., Guo, Y., Xu, T., & Yuan, X. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-59861-7_33 · Cited by 3 · see citing works
Weakly-Supervised 2D/3D Image Registration via Differentiable X-ray Rendering and ROI Segmentation
Cui, Y., Meng, M. Q. ‐., & Min, Z. (2025) · Lecture notes in computer science · doi:10.1007/978-3-032-04984-1_62 · Cited by 2 · see citing works
Thibeault, S., Roy‐Beaudry, M., Stefan Parent, D., & Kadoury, S. (2024) · Medical Image Analysis · doi:10.1016/j.media.2024.103400 · Cited by 2 · see citing works
Artificial Intelligence in Hip Arthroplasty
Zampogna, B., Parisi, F. R., Zampoli, A., et al. (2024) · Springer eBooks · doi:10.1007/978-3-031-70310-2_5 · Cited by 1 · see citing works
Rapid patient-specific neural networks for intraoperative X-ray to volume registration
Gopalakrishnan, V., Chlorogiannis, D.-D., Abumoussa, A., Larson, A. M., Haouchine, N., Orbach, D. B., Frisken, S., Dey, N., & Golland, P. (2025) · arXiv · doi:10.48550/arxiv.2503.16309 · Cited by 2 · see citing works
Toward real-time alignment of 3D CT and 2D X-ray with multi-stage CNNs
Facente, F., Billot, B., Gopalakrishnan, V., Kattel, M., Wei, W., Golland, P., Delingette, H., Ayache, N., & Berthet-Rayne, P. (2026) · Computer Assisted Surgery · doi:10.1080/24699322.2026.2684372 · Cited by 1 · see citing works
Shen, A., Fu, X., Jiang, J., Zeng, Q., Tang, Y., Chen, Z., Nong, L., Wang, F., & Zhou, S. K. (2025) · Lecture notes in computer science · doi:10.1007/978-3-032-05114-1_44 · Cited by 1 · see citing works
Wang, J., Vogl, F., Schütz, P., et al. (2025) · Lecture notes in computer science · doi:10.1007/978-3-032-05169-1_62 · Cited by 1 · see citing works
“Existing single-plane fluoroscopic datasets provide accurate pose labels for hip anatomy (DeepFluoro) [6] and for knee implants in total knee arthroplasty (CAMS-Knee, Stan) [18,2] where precise kinematics are available.”
Towards Seamless Integration of Magnetic Tracking Into Fluoroscopy-Guided Interventions
Xing, S., Mirzaei, M., Xia, W., Ahmed-Fazal, I., Pardasani, U., Jayarathne, U. L., Illsley, S., Cardarelli‐Leite, L., Fenster, A., Peters, T. M., et al. (2025) · IEEE Transactions on Biomedical Engineering · doi:10.1109/tbme.2025.3596877 · Cited by 2 · see citing works
“3) Intensity-based registration: Beginning with the initial C-arm pose, an image intensity-based multi-resolution registration strategy [26], [27] was employed to optimize the patient pose parameters (i.e., T sourcepatient ).”
Uncertainty Quantification in Image-based 2D/3D Registration and Its Relationship with Accuracy
Cho, S. M., Do, A., Grupp, R. B., Armand, M., Taylor, R. H., & Unberath, M. (2025) · International Journal of Computer Assisted Radiology and Surgery · doi:10.1007/s11548-025-03417-x · Cited by 1 · see citing works
Self-citationOu, D., & Xiao, Y. (2025) · Journal of Nondestructive Evaluation · doi:10.1007/s10921-025-01191-z · Cited by 1 · see citing works
Burton, W. S., Myers, C. A., Clary, C. W., et al. (2025) · Medical Engineering & Physics · doi:10.1016/j.medengphy.2025.104351 · Cited by 1 · see citing works
Registration of 2D X-ray and 3D CT Images for Robot-Assisted Pelvic Fracture Reduction Surgery
Jiang, P., Lü, S., Yang, Y., et al. (2024) · IEEE International Conference on Robotics and Biomimetics (ROBIO) · doi:10.1109/robio64047.2024.10907299 · Cited by 1 · see citing works
Chiang, L.-H., Chen, Y.-C., Huang, G., et al. (2023) · Quantitative Imaging in Medicine and Surgery · doi:10.21037/qims-23-490 · Cited by 1 · see citing works
FlowReg: Reconstruction-Guided Flow Matching for Universal Spinal CT/X-Ray Registration
Shen, A., Jiang, J., Fu, X., Zeng, Q., Tang, Y., Chen, Z., Nong, L., Wang, F., & Zhou, S. K. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-38186-6_8
Intraoperative Fluoroscopy-Volume Registration Using Virtual-Supervision Skeleton Attention
Zhang, S., Feng, N., Fan, W., et al. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-38186-6_13
Rapid patient-specific neural networks for X-ray to volume registration
Gopalakrishnan, V., Chlorogiannis, D., Abumoussa, A. L., Larson, A. M., Haouchine, N., Orbach, D. B., Frisken, S., Dey, N., & Golland, P. (2026) · Nature · doi:10.1038/s41586-026-11045-x · Cited by 7 · see citing works
Jiang, J., Shen, A., Wang, Z., Fu, X., ZhengMing, C. H. E. N., Nong, L., Wang, F., & Zhou, S. K. (2026) · Medical Image Analysis · doi:10.1016/j.media.2026.104319
Meta-reg: A meta-learning framework for generalizable intraoperative 2D/3D image registration
Xu, L., Kong, Q., Wang, X., Zhang, Y., & Xu, L. (2026) · Applied Soft Computing · doi:10.1016/j.asoc.2026.116093
Wang, J., Li, X., Surbeck, R., et al. (2026) · International Journal of Computer Assisted Radiology and Surgery · doi:10.1007/s11548-026-03734-9
Wang, Q., Liu, P., Cai, S., & Zhang, L. (2026) · Biomimetic Intelligence and Robotics · doi:10.1016/j.birob.2026.100349
Wu, L., Seibold, M., Cavalcanti, N., et al. (2026) · Medical Image Analysis · doi:10.1016/j.media.2026.104133
“Here, intraoperative imaging is utilized for surface reconstruction, such as intraoperative CT, X-rays (Grupp et al., 2020; Ku et al., 2025; Leung et al., 2010) and ultrasound (Wein et al., 2015; Salehi et al., 2017; Eppenhof et al., 2019; Guo et al., 2021; Ciganovic et al., 2018-04-05; Fanti et al., 2018; Li et al.,…”
PixPose: End-to-end 2D/3D medical image registration via image translation
Wu, K., Zhang, W., Zhang, X., Yin, Y., & Wang, R. (2026) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2026.110312
Jiang, P., Lei, L., Lu, S., et al. (2026) · IEEE International Symposium on Biomedical Imaging (ISBI) · doi:10.1109/isbi61048.2026.11515851
“This study was conducted retrospectively using publicly available human subject data from the open-access datasets [10].”
Chen, M., & Kong, Y. (2026) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv61042.2026.00699
“We perform experiments on three publicly available datasets to evaluate the effectiveness of the proposed method: DeepFluoro [22], Ljubljana [48], and CT-Spine1K [10].”
Semantic edge‐guided single‐view 2D/3D registration for vertebrae in X‐rays
Shen, A., Jiang, J., Tang, Y., ZhengMing, C. H. E. N., Nong, L., Chen, Q., Wang, F., & Wu, X. (2026) · Medical Physics · doi:10.1002/mp.70385 · Cited by 1 · see citing works
“In single vertebra registration on simulated data, we compared the proposed method with three representative methods: the gradient-based method (Xreg), 14 Pro-ST, 43 and PSSS. 16 We found that the proposed method redsubstantially outperforms these benchmarks in terms of registration accuracy and efficiency on the Test…”
InfluentialMulti-stage CNN for Fast Registration of 3D Preoperative CTs to 2D Intraoperative X-Rays
Facente, F., Billot, B., Gopalakrishnan, V., Kattel, M., Wei, W., Golland, P., Delingette, H., Ayache, N., & Berthet-Rayne, P. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-09784-2_14
Burton, W. S., Myers, C. A., & Rullkoetter, P. J. (2025) · Annals of Biomedical Engineering · doi:10.1007/s10439-025-03911-y
Yu, W., Guo, X., Liu, X. Z., Liu, Y., Zheng, H., Huang, Y., & Yuan, Y. (2025) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51701.2025.01995
“1 illustrates the quantitative results of our Gaus-sianReg and other redistration methods on DeepFluoro dataset, including regression-based methods (PoseNet [18], Deep6DPose [9], GDR-Net [42], DiffPose [13]) and landmark-based methods (FixedLandmark [14], RayEmb [37]).”
InfluentialSuh, Y., Martin, J. R., & Moyer, D. (2025) · Lecture notes in computer science · doi:10.1007/978-3-032-05479-1_1
Ku, P., Liu, M., Grupp, R. B., et al. (2025) · International Journal of Computer Assisted Radiology and Surgery · doi:10.1007/s11548-025-03426-w
Self-citationOu, D., & Xiao, Y. (2025) · Research Square · doi:10.21203/rs.3.rs-5831887/v1
Reliable 2D–3D Registration in Dynamic Stereo-Radiography With Energy Barrier Constraints
Burton, W. S., Myers, C. A., Clary, C. W., & Rullkoetter, P. J. (2024) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2024.3522200
“A composite objective function was optimized, consisting of pNCC and a regularization term which penalized differences between CNN-predicted key points and projected anatomic landmarks [14].”
InfluentialBurton, W. S., Gibbons, K. D., & Economon, K. (2024) · Computer Methods in Biomechanics and Biomedical Engineering Imaging & Visualization · doi:10.1080/21681163.2024.2415716
“…models reconstructed from PS scans are leveraged for surgical planning or guidance (Lombardi and Berend 2008; Fadero and Shah 2014; Verborgt et al. 2014; Wong 2016; Lutz et al. 2019; Grupp et al. 2020; Fan et al. 2022), and for creation of PS instrumentation (Nam et al. 2012; Mattei et al. 2016).”
Zhou, C., Faruqui, S. H. A., An, D., Patel, A., Abdalla, R. N., Hurley, M. C., Shaibani, A., Potts, M. B., Jahromi, B. S., Ansari, S. A., et al. (2024) · Journal of Imaging Informatics in Medicine · doi:10.48550/arxiv.2308.00214 · Cited by 1 · see citing works
Techniques for automatic landmark detection of human tissue
Jiménez-Perez, J. R., Negrillo-Cárdenas, J., & Feito, F. R. (2023) · Elsevier eBooks · doi:10.1016/b978-0-12-823913-1.00005-1
- Authors
- Grupp, R. B., Unberath, M., Gao, C., Hegeman, R. A., Murphy, R. J., Alexander, C. P., Otake, Y., McArthur, B. A., Armand, M., & Taylor, R. H.
- Affiliations
- Johns Hopkins University (US)
- Johns Hopkins University Applied Physics Laboratory (US)
- Auris Health (United States) (US)
- Nara Institute of Science and Technology (JP)
- The University of Texas at Austin (US)
- Corresponding author
- Grupp, Robert B.
- ORCID
- Venue
- International Journal of Computer Assisted Radiology and Surgery
- Published
- 2020-04-24
- Volume
- 15
- Issue
- 5
- Pages
- 759-769
- ISSN
- 1861-6410 (ISSN-L)
- 1861-6429
- arXiv
- 1911.07042
- PMC
- PMC7263976
- OpenAlex
- W3017446891
- Cited by
- 81
- References
- 44
- Citations by year
- 2026: 14 · 2025: 15 · 2024: 16 · 2023: 16 · 2022: 4 · 2021: 9 · 2020: 6
- FWCI
- 6.94
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- International Journal of Computer Assisted Radiology and Surgery (publishedVersion)
- PubMed (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- PubMed Central (submittedVersion)
- arXiv (Cornell University)
- Funders
- National Institute of Biomedical Imaging and Bioengineering (R01EB006839, R21EB020113)
- Johns Hopkins University Applied Physics Laboratory (FNACCX24)
- Topics
- Health Sciences › Medicine › Surgery › Orthopaedic implants and arthroplasty
- Health Sciences › Medicine › Surgery › Pelvic and Acetabular Injuries
- Health Sciences › Medicine › Radiology, Nuclear Medicine and Imaging › Medical Imaging Techniques and Applications
- Keywords
- 2D-3D registration, intraoperative registration, computer-assisted navigation, hip surgery, neural network segmentation, anatomical landmark detection, femur segmentation, annotated dataset
- MeSH terms
- Algorithms, Femur, Fluoroscopy, Humans, Pelvis
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-09-29.
Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-09-29.
Cited sources are checked by gpt-6-sol.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
