International Journal of Computer Assisted Radiology and Surgery最新文献

{"title":"Point cloud registration algorithm using liver vascular skeleton feature with computed tomography and ultrasonography image fusion.","authors":"Satoshi Miura, Masayuki Nakayama, Kexin Xu, Zhang Bo, Ryoko Kuromatsu, Masahito Nakano, Yu Noda, Takumi Kawaguchi","doi":"10.1007/s11548-025-03496-w","DOIUrl":"https://doi.org/10.1007/s11548-025-03496-w","url":null,"abstract":"<p><strong>Purpose: </strong>Radiofrequency ablation for liver cancer has advanced rapidly. For accurate ultrasound-guided soft-tissue puncture surgery, it is necessary to fuse intraoperative ultrasound images with preoperative computed tomography images. However, the conventional method is difficult to estimate and fuse images accurately. To address this issue, the present study proposes an algorithm for registering cross-source point clouds based on not surface but the geometric features of the vascular point cloud.</p><p><strong>Methods: </strong>We developed a fusion system that performs cross-source point cloud registration between ultrasound and computed tomography images, extracting the node, skeleton, and geomatic feature of the vascular point cloud. The system completes the fusion process in an average of 14.5 s after acquiring the vascular point clouds via ultrasound.</p><p><strong>Results: </strong>The experiments were conducted to fuse liver images by the dummy model and the healthy participants, respectively. The results show the proposed method achieved a registration error within 1.4 mm and decreased the target registration error significantly compared to other methods in a liver dummy model registration experiment. Furthermore, the proposed method achieved the averaged RMSE within 2.23 mm in a human liver vascular skeleton.</p><p><strong>Conclusion: </strong>The study concluded that because the registration method using vascular feature point cloud could realize the rapid and accurate fusion between ultrasound and computed tomography images, the method is useful to apply the real puncture surgery for radiofrequency ablation for liver. In future work, we will evaluate the proposed method by the patients.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-08-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144977838","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The interpretable surgical temporal informer: explainable surgical time completion prediction.","authors":"Roger D Soberanis-Mukul, Rohit Shankar, Lalithkumar Seenivasan, Jose L Porras, Masaru Ishii, Mathias Unberath","doi":"10.1007/s11548-025-03448-4","DOIUrl":"https://doi.org/10.1007/s11548-025-03448-4","url":null,"abstract":"<p><strong>Purpose: </strong>Predicting surgical time completion helps streamline surgical workflow and OR utilization, enhancing hospital efficacy. When time prediction is based on interventional video of the surgical site, time predictions may correlate with technical proficiency of the surgeon because skill is a useful proxy of completion time. To understand features that are predictive of surgical time in surgical site video, we develop prototype-like visual explanations, making them applicable to video sequences.</p><p><strong>Methods: </strong>We introduce an interpretable method for predicting surgical duration by identifying prototype-like patterns within egocentric video of the surgical site. Unlike conventional image-based prototype models that generate patch-based prototypes, our method extracts video-based explanations tied to segments of surgical videos with similar time deviation patterns. We achieve this by comparing the principal components of feature representation differences at various time points in the predictions. To effectively capture long-range dependencies in the prediction task, we employ an informer as the primary predictive model.</p><p><strong>Results: </strong>This model is applied to a dataset of 42 point-of-view craniotomy videos, collected under an approved IRB protocol. On average, our interpretable model performs better than the baseline models in surgical time completion.</p><p><strong>Conclusion: </strong>Our approach not only contributes to the interpretability of surgical time predictions but also takes full advantage of the detailed information provided by surgical video data.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-08-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144977807","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"RANSAC-based global 3DUS to CT/MR rigid registration using liver surface and vessels.","authors":"Tsubasa Goto, Riki Igarashi, Iku Cho, Kazushi Numata, Yugo Ishino, Yoshiro Kitamura, Masafumi Noguchi, Takanori Hirai, Koji Waki","doi":"10.1007/s11548-025-03498-8","DOIUrl":"https://doi.org/10.1007/s11548-025-03498-8","url":null,"abstract":"<p><strong>Purpose: </strong>Fusion imaging requires initial registration of ultrasound (US) images using computed tomography (CT) or magnetic resonance (MR) imaging. The sweep position of US depends on the procedure. For instance, the liver may be observed in intercostal, subcostal, or epigastric positions. However, no well-established method for automatic initial registration accommodates all positions. A global rigid 3D-3D registration technique aimed at developing an automatic registration method independent of the US sweep position is proposed.</p><p><strong>Methods: </strong>The proposed technique utilizes the liver surface and vessels, such as the portal and hepatic veins, as landmarks. The algorithm segments the liver region and vessels from both US and CT/MR images using deep learning models. Based on these outputs, the point clouds of the liver surface and vessel centerlines were extracted. The rigid transformation parameters were estimated through point cloud registration using a RANSAC-based algorithm. To enhance speed and robustness, the RANSAC procedure incorporated constraints regarding the possible ranges for each registration parameter based on the relative position and orientation of the probe and body surface.</p><p><strong>Results: </strong>Registration accuracy was quantitatively evaluated using clinical data from 80 patients, including US images taken from the intercostal, subcostal, and epigastric regions. The registration errors were 7.3 ± 3.2, 9.3 ± 3.7, and 8.4 ± 3.9 mm for the intercostal, subcostal, and epigastric regions, respectively.</p><p><strong>Conclusion: </strong>The proposed global rigid registration technique fully automated the complex manual registration required for liver fusion imaging and enhanced the workflow efficiency of physicians and sonographers.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-08-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144977800","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Time series-derived fractal dimension of CT perfusion in acute ischemic stroke: a promising marker for hypoperfused tissue quantification.","authors":"Shota Ichikawa, Yohan Kondo, Satoshi Yokoyama","doi":"10.1007/s11548-025-03500-3","DOIUrl":"https://doi.org/10.1007/s11548-025-03500-3","url":null,"abstract":"<p><strong>Purpose: </strong>Computed tomography perfusion (CTP) imaging for acute ischemic stroke relies on accurately identifying hypoperfused brain tissue to guide treatment decisions. However, deconvolution-based methods often suffer from variability in perfusion parameters and lesion volumes across different software. This study evaluated the feasibility of temporal fractal analysis, specifically, time series-derived fractal dimension (FD) using the Higuchi method, as a biomarker for detecting hypoperfused brain tissue.</p><p><strong>Methods: </strong>Fractal analysis was applied to voxel-wise time-series data from both simulated phantom datasets and 149 CTP images from the publicly available Ischemic Stroke Lesion Segmentation (ISLES) 2024 dataset. FD was calculated using optimized parameters determined through the phantom study. In the patient study, the ischemic core was defined by follow-up MRI, and the penumbra was defined as tissue with Tmax > 6 s. FD values were statistically compared between core, penumbra, and normal tissue. Diagnostic performance was assessed using receiver operating characteristic (ROC) analysis.</p><p><strong>Results: </strong>In the phantom study, FD showed a strong correlation (ρ > 0.9) with true cerebral blood flow (CBF) across all cerebral blood volume (CBV) values when the tuning parameter k_max was optimized based on the number of CTP frames. In the patient study, FD differed significantly across tissue types (p < 0.001). For penumbra versus normal classification, FD achieved an AUC of 0.732, outperforming CBF and CBV (p < 0.001). In core versus penumbra classification, FD showed the highest AUC of 0.641 among all metrics.</p><p><strong>Conclusion: </strong>Time series-derived FD offers a promising approach to characterizing perfusion abnormalities in stroke, with potential as a complementary metric to conventional CTP parameters.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144876624","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Interactive AI annotation of medical images in a virtual reality environment.","authors":"Lotta Orsmaa, Mikko Saukkoriipi, Jari Kangas, Nastaran Rasouli, Jorma Järnstedt, Helena Mehtonen, Jaakko Sahlsten, Joel Jaskari, Kimmo Kaski, Roope Raisamo","doi":"10.1007/s11548-025-03497-9","DOIUrl":"https://doi.org/10.1007/s11548-025-03497-9","url":null,"abstract":"<p><strong>Purpose: </strong>Artificial intelligence (AI) achieves high-quality annotations of radiological images, yet often lacks the robustness required in clinical practice. Interactive annotation starts with an AI-generated delineation, allowing radiologists to refine it with feedback, potentially improving precision and reliability. These techniques have been explored in two-dimensional desktop environments, but are not validated by radiologists or integrated with immersive visualization technologies. We used a Virtual Reality (VR) system to determine whether (1) the annotation quality improves when radiologists can edit the AI annotation and (2) whether the extra work done by editing is worthwhile.</p><p><strong>Methods: </strong>We evaluated the clinical feasibility of an interactive VR approach to annotate mandibular and mental foramina on segmented 3D mandibular models. Three experienced dentomaxillofacial radiologists reviewed AI-generated annotations and, when needed, refined them at the voxel level in 3D space through click-based interactions until clinical standards were met.</p><p><strong>Results: </strong>Our results indicate that integrating expert feedback within an immersive VR environment enhances annotation accuracy, improves clinical usability, and offers valuable insights for developing medical image analysis systems incorporating radiologist input.</p><p><strong>Conclusion: </strong>This study is the first to compare the quality of original and interactive AI annotation and to use radiologists' opinions as the measure. More research is needed for generalization.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144876623","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Toward ICE-XRF fusion: real-time pose estimation of the intracardiac echo probe in 2D X-ray using deep learning.","authors":"Annelies Severens, Midas Meijs, Vipul Pai Raikar, Richard Lopata","doi":"10.1007/s11548-025-03493-z","DOIUrl":"https://doi.org/10.1007/s11548-025-03493-z","url":null,"abstract":"<p><strong>Purpose: </strong>Valvular heart disease affects 2.5% of the general population and 10% of people aged over 75, with many patients untreated due to high surgical risks. Transcatheter valve therapies offer a safer, less invasive alternative but rely on ultrasound and X-ray image guidance. The current ultrasound technique for valve interventions, transesophageal echocardiography (TEE), requires general anesthesia and has poor visibility of the right side of the heart. Intracardiac echocardiography (ICE) provides improved 3D imaging without the need for general anesthesia but faces challenges in adoption due to device handling and operator training.</p><p><strong>Methods: </strong>To facilitate the use of ICE in the clinic, the fusion of ultrasound and X-ray is proposed. This study introduces a two-stage detection algorithm using deep learning to support ICE-XRF fusion. Initially, the ICE probe is coarsely detected using an object detection network. This is followed by 5-degree-of-freedom (DoF) pose estimation of the ICE probe using a regression network.</p><p><strong>Results: </strong>Model validation using synthetic data and seven clinical cases showed that the framework provides accurate probe detection and 5-DoF pose estimation. For the object detection, an F1 score of 1.00 was achieved on synthetic data and high precision (0.97) and recall (0.83) for clinical cases. For the 5-DoF pose estimation, median position errors were found under 0.5mm and median rotation errors below <math><mrow><mn>7</mn> <mo>.</mo> <msup><mn>2</mn> <mo>∘</mo></msup> </mrow> </math> .</p><p><strong>Conclusion: </strong>This real-time detection method supports image fusion of ICE and XRF during clinical procedures and facilitates the use of ICE in valve therapy.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144876625","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Benchmarking NousNav: quantifying the spatial accuracy and clinical performance of an affordable, open-source neuronavigation system.","authors":"Colton Barr, Colin Galvin, Parikshit Juvekar, Erickson Torio, Samantha Horvath, Samantha Sadler, Annie Li, Ryan Bardsley, Tina Kapur, Steve Pieper, Sonia Pujol, Sarah Frisken, Gabor Fichtinger, Alexandra Golby","doi":"10.1007/s11548-025-03494-y","DOIUrl":"10.1007/s11548-025-03494-y","url":null,"abstract":"<p><strong>Purpose: </strong>NousNav is a low-cost, open-source neuronavigation platform built to address the high costs and resource limitations that hinder access to advanced neurosurgical technologies in low-resource settings. The low-cost and accessibility of the system is made possible using consumer-grade optical tracking and open-source software packages. This study aims to assess the performance of these core enabling technologies by quantifying their spatial accuracy and comparing it to a commercial gold standard.</p><p><strong>Methods: </strong>A series of experiments were conducted to evaluate the capabilities of the selected hardware and registration infrastructure utilized in NousNav. Each component was tested both in a simulated bench-top environment and clinically across four brain tumor resection cases.</p><p><strong>Results: </strong>The Optitrack Duo tracker used by NousNav was found to have a mean localization error of 0.8mm (SD 0.4mm). In bench-top phantom testing, NousNav had an average target registration error of 5.0mm (SD 2.3mm) following patient registration. Clinical evaluations revealed a mean distance of 4.2mm (SD 1.5mm) between points reported by NousNav versus those obtained using a commercial neuronavigation system.</p><p><strong>Conclusion: </strong>These experiments highlight the role of baseline camera tracking performance, tracked instrument calibration, and patient positioning on the spatial performance of NousNav. They also provide an essential benchmark assessment of the system to help inform future clinical use-cases and direct ongoing system development.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-08-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12469177/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144856940","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A multi-view validation framework for LLM-generated knowledge graphs of chronic kidney disease.","authors":"Aditya Kumar, Dilpreet Singh, Mario Cypko, Oliver Amft","doi":"10.1007/s11548-025-03495-x","DOIUrl":"https://doi.org/10.1007/s11548-025-03495-x","url":null,"abstract":"<p><strong>Purpose: </strong>The goal of our work is to develop a multi-view validation framework for evaluating LLM-generated knowledge graph (KG) triples. The proposed approach aims to address the lack of established validation procedure in the context of LLM-supported KG construction.</p><p><strong>Methods: </strong>The proposed framework evaluates the LLM-generated triples across three dimensions: semantic plausibility, ontology-grounded type compatibility, and structural importance. We demonstrate the performance for GPT-4 generated concept-specific (e.g., for medications, diagnosis, procedures) triples in the context of chronic kidney disease (CKD).</p><p><strong>Results: </strong>The proposed approach consistently achieves high-quality results across evaluated GPT-4 generated triples, strong semantic plausibility (semantic score mean: 0.79), excellent type compatibility (type score mean: 0.84), and high structural importance of entities within the CKD knowledge domain (ResourceRank mean: 0.94).</p><p><strong>Conclusion: </strong>The validation framework offers a reliable and scalable method for evaluating quality and validity of LLM-generated triples across three views: semantic plausibility, type compatibility, and structural importance. The framework demonstrates robust performance in filtering high-quality triples and lays a strong foundation for fast and reliable medical KG construction and validation.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-08-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144856939","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Semi-automatic puncture robotic system based on real-time multi-modal image fusion: preclinical evaluation.","authors":"Bo Zhang, Kui Chen, Yuhang Yao, Bo Wu, Qiang Li, Zheming Zhang, Peihua Fan, Wei Wang, Manxia Lin, Xiang Jing, Shigeki Sugano, Masakatsu G Fujie, Ming Kuang","doi":"10.1007/s11548-025-03471-5","DOIUrl":"10.1007/s11548-025-03471-5","url":null,"abstract":"<p><strong>Purpose: </strong>Traditional surgical robot system relying on computed tomography (CT) navigation suffers from two drawbacks during abdominal organ puncture surgeries. Firstly, the puncture target is displaced under the influence of respiration, thereby reducing the puncture accuracy. Secondly, the puncture process lacks real-time visualization, which may potentially give rise to medical accidents. This paper presents a semi-automatic surgical robot system based on the fusion guidance of real-time ultrasound images and CT, along with the monitoring of the patient's respiratory state, to address these issues.</p><p><strong>Method: </strong>This system utilizes a six-axis force sensor in contact with the human body, and a respiratory model can be constructed through data got from force sensor to monitor the patient's respiratory phase and recommend the optimal puncture phase. The issue of non-real-time puncture guidance is addressed through the real-time registration and fusion of ultrasound (US) images with preoperative CT images.</p><p><strong>Results: </strong>Phantom experiments and animal experiments were carried out based on this design. The test results indicate that in these two experiments, the average fusion error between US and CT of the main tissues in the liver is within 3 mm. For puncture accuracy, in the phantom experiment, the average puncture error was 1.0 mm, with a minimum of 0 mm and a maximum of 2.1 mm. In the animal experiment, the average puncture error was 2.5 mm, ranging from a minimum of 1.6 mm to a maximum of 3.0 mm.</p><p><strong>Conclusion: </strong>The results of two experiments show both the image fusion accuracy and puncture accuracy of this system are within 3mm, which can meet the requirement of 5 mm puncture accuracy in clinical practice. Approximately 70% of the operation are automatically accomplished by robot system, greatly reducing the reliance on the doctor's experience.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-08-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144856941","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Towards a zero-shot low-latency navigation for open surgery augmented reality applications.","authors":"Michael Schwimmbeck, Serouj Khajarian, Christopher Auer, Thomas Wittenberg, Stefanie Remmele","doi":"10.1007/s11548-025-03480-4","DOIUrl":"https://doi.org/10.1007/s11548-025-03480-4","url":null,"abstract":"<p><strong>Purpose: </strong>Augmented reality (AR) enhances surgical navigation by superimposing visible anatomical structures with three-dimensional virtual models using head-mounted displays (HMDs). In particular, interventions such as open liver surgery can benefit from AR navigation, as it aids in identifying and distinguishing tumors and risk structures. However, there is a lack of automatic and markerless methods that are robust against real-world challenges, such as partial occlusion and organ motion.</p><p><strong>Methods: </strong>We introduce a novel multi-device approach for automatic live navigation in open liver surgery that enhances the visualization and interaction capabilities of a HoloLens 2 HMD through precise and reliable registration using an Intel RealSense RGB-D camera. The intraoperative RGB-D segmentation and the preoperative CT data are utilized to register a virtual liver model to the target anatomy. An AR-prompted Segment Anything Model (SAM) enables robust segmentation of the liver in situ without the need for additional training data. To mitigate algorithmic latency, Double Exponential Smoothing (DES) is applied to forecast registration results.</p><p><strong>Results: </strong>We conducted a phantom study for open liver surgery, investigating various scenarios of liver motion, viewpoints, and occlusion. The mean registration errors (8.31 mm-18.78 mm TRE) are comparable to those reported in prior work, while our approach demonstrates high success rates even for high occlusion factors and strong motion. Using forecasting, we bypassed the algorithmic latency of 79.8 ms per frame, with median forecasting errors below 2 mms and 1.5 degrees between the quaternions.</p><p><strong>Conclusion: </strong>To our knowledge, this is the first work to approach markerless in situ visualization by combining a multi-device method with forecasting and a foundation model for segmentation and tracking. This enables a more reliable and precise AR registration of surgical targets with low latency. Our approach can be applied to other surgical applications and AR hardware with minimal effort.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-08-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144790679","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}