Jorge Badilla-Solórzano, Sontje Ihler, Thomas Seel
{"title":"HybGrip: a synergistic hybrid gripper for enhanced robotic surgical instrument grasping.","authors":"Jorge Badilla-Solórzano, Sontje Ihler, Thomas Seel","doi":"10.1007/s11548-024-03245-5","DOIUrl":"https://doi.org/10.1007/s11548-024-03245-5","url":null,"abstract":"<p><strong>Purpose: </strong>A fundamental task of a robotic scrub nurse is handling surgical instruments. Thus, a gripper capable of consistently grasping a wide variety of tools is essential. We introduce a novel gripper that combines granular jamming and pinching technologies to achieve a synergistic improvement in surgical instrument grasping.</p><p><strong>Methods: </strong>A reliable hybrid gripper is constructed by integrating a pinching mechanism and a standard granular jamming gripper, achieving enhanced granular interlocking. For our experiments, our prototype is affixed to the end-effector of a collaborative robot. A novel grasping strategy is proposed and utilized to evaluate the robustness and performance of our prototype on 18 different surgical tools with diverse geometries.</p><p><strong>Results: </strong>It is demonstrated that the integration of the pinching mechanism significantly enhances grasping performance compared with standard granular jamming grippers, with a success rate above 98%. It is shown that with the combined use of our gripper with an underlying grid, i.e., a complementary device placed beneath the instruments, robustness and performance are further enhanced.</p><p><strong>Conclusion: </strong>Our prototype's performance in surgical instrument grasping stands on par with, if not surpasses, that of comparable contemporary studies, ensuring its competitiveness. Our gripper proves to be robust, cost-effective, and simple, requiring no instrument-specific grasping strategies. Future research will focus on addressing the sterilizability of our prototype and assessing the viability of the introduced grid for intra-operative use.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":null,"pages":null},"PeriodicalIF":2.3,"publicationDate":"2024-08-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142019487","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":"Artificial intelligence facilitates the potential of simulator training: An innovative laparoscopic surgical skill validation system using artificial intelligence technology.","authors":"Atsuhisa Fukuta, Shogo Yamashita, Junnosuke Maniwa, Akihiko Tamaki, Takuya Kondo, Naonori Kawakubo, Kouji Nagata, Toshiharu Matsuura, Tatsuro Tajiri","doi":"10.1007/s11548-024-03253-5","DOIUrl":"https://doi.org/10.1007/s11548-024-03253-5","url":null,"abstract":"<p><strong>Purpose: </strong>The development of innovative solutions, such as simulator training and artificial intelligence (AI)-powered tutoring systems, has significantly changed surgical trainees' environments to receive the intraoperative instruction necessary for skill acquisition. In this study, we developed a new objective assessment system using AI for forceps manipulation in a surgical training simulator.</p><p><strong>Methods: </strong>Laparoscopic exercises were recorded using an iPad®, which provided top and side views. Top-view movies were used for AI learning of forceps trajectory. Side-view movies were used as supplementary information to assess the situation. We used an AI-based posture estimation method, DeepLabCut (DLC), to recognize and positionally measure the forceps in the operating field. Tracking accuracy was quantitatively evaluated by calculating the pixel differences between the annotation points and the points predicted by the AI model. Tracking stability at specified key points was verified to assess the AI model.</p><p><strong>Results: </strong>We selected a random sample to evaluate tracking accuracy quantitatively. This sample comprised 5% of the frames not used for AI training from the complete set of video frames. We compared the AI detection positions and correct positions and found an average pixel discrepancy of 9.2. The qualitative evaluation of the tracking stability was good at the forceps hinge; however, forceps tip tracking was unstable during rotation.</p><p><strong>Conclusion: </strong>The AI-based forceps tracking system can visualize and evaluate laparoscopic surgical skills. Improvements in the proposed system and AI self-learning are expected to enable it to distinguish the techniques of expert and novice surgeons accurately. This system is a useful tool for surgeon training and assessment.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":null,"pages":null},"PeriodicalIF":2.3,"publicationDate":"2024-08-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142005803","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}
Mattia Magro, Nicola Covallero, Elena Gambaro, Emanuele Ruffaldi, Elena De Momi
{"title":"A dual-instrument Kalman-based tracker to enhance robustness of microsurgical tools tracking.","authors":"Mattia Magro, Nicola Covallero, Elena Gambaro, Emanuele Ruffaldi, Elena De Momi","doi":"10.1007/s11548-024-03246-4","DOIUrl":"https://doi.org/10.1007/s11548-024-03246-4","url":null,"abstract":"<p><strong>Purpose: </strong>The integration of a surgical robotic instrument tracking module within optical microscopes holds the potential to advance microsurgery practices, as it facilitates automated camera movements, thereby augmenting the surgeon's capability in executing surgical procedures.</p><p><strong>Methods: </strong>In the present work, an innovative detection backbone based on spatial attention module is implemented to enhance the detection accuracy of small objects within the image. Additionally, we have introduced a robust data association technique, capable to re-track surgical instrument, mainly based on the knowledge of the dual-instrument robotics system, Intersection over Union metric and Kalman filter.</p><p><strong>Results: </strong>The effectiveness of this pipeline was evaluated through testing on a dataset comprising ten manually annotated videos of anastomosis procedures involving either animal or phantom vessels, exploiting the Symani®Surgical System-a dedicated robotic platform designed for microsurgery. The multiple object tracking precision (MOTP) and the multiple object tracking accuracy (MOTA) are used to evaluate the performance of the proposed approach, and a new metric is computed to demonstrate the efficacy in stabilizing the tracking result along the video frames. An average MOTP of 74±0.06% and a MOTA of 99±0.03% over the test videos were found.</p><p><strong>Conclusion: </strong>These results confirm the potential of the proposed approach in enhancing precision and reliability in microsurgical instrument tracking. Thus, the integration of attention mechanisms and a tailored data association module could be a solid base for automatizing the motion of optical microscopes.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":null,"pages":null},"PeriodicalIF":2.3,"publicationDate":"2024-08-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141918025","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}
Luca Wegener, Dirk Wilhelm, Maximilian Berlet, Jonas Fuchtmann
{"title":"Development of a human machine interface for robotically assisted surgery optimized for laparoscopic workflows.","authors":"Luca Wegener, Dirk Wilhelm, Maximilian Berlet, Jonas Fuchtmann","doi":"10.1007/s11548-024-03239-3","DOIUrl":"https://doi.org/10.1007/s11548-024-03239-3","url":null,"abstract":"<p><strong>Introduction: </strong>In robotic-assisted surgery (RAS), the input device is the primary site for the flow of information between the user and the robot. Most RAS systems remove the surgeon's console from the sterile surgical site. Beneficial for performing lengthy procedures with complex systems, this ultimately lacks the flexibility that comes with the surgeon being able to remain at the sterile site.</p><p><strong>Methods: </strong>A prototype of an input device for RAS is constructed. The focus lies on intuitive control for surgeons and a seamless integration into the surgical workflow within the sterile environment. The kinematic design is translated from the kinematics of laparoscopic surgery. The input device uses three degrees of freedom from a flexible instrument as input. The prototype's performance is compared to that of a commercially available device in an evaluation. Metrics are used to evaluate the surgeons' performance with the respective input device in a virtual environment implemented for the evaluation.</p><p><strong>Results: </strong>The evaluation of the two input devices shows statistically significant differences in the performance metrics. With the proposed prototype, the surgeons perform the tasks faster, more precisely, and with fewer errors.</p><p><strong>Conclusion: </strong>The prototype is an efficient and intuitive input device for surgeons with laparoscopic experience. The placement in the sterile working area allows for seamless integration into the surgical workflow and can potentially enable new robotic approaches.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":null,"pages":null},"PeriodicalIF":2.3,"publicationDate":"2024-08-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141914558","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":"Adaptive neighborhood triplet loss: enhanced segmentation of dermoscopy datasets by mining pixel information.","authors":"Mohan Xu, Lena Wiese","doi":"10.1007/s11548-024-03241-9","DOIUrl":"https://doi.org/10.1007/s11548-024-03241-9","url":null,"abstract":"<p><strong>Purpose: </strong>The integration of deep learning in image segmentation technology markedly improves the automation capabilities of medical diagnostic systems, reducing the dependence on the clinical expertise of medical professionals. However, the accuracy of image segmentation is still impacted by various interference factors encountered during image acquisition.</p><p><strong>Methods: </strong>To address this challenge, this paper proposes a loss function designed to mine specific pixel information which dynamically changes during training process. Based on the triplet concept, this dynamic change is leveraged to drive the predicted boundaries of images closer to the real boundaries.</p><p><strong>Results: </strong>Extensive experiments on the PH2 and ISIC2017 dermoscopy datasets validate that our proposed loss function overcomes the limitations of traditional triplet loss methods in image segmentation applications. This loss function not only enhances Jaccard indices of neural networks by 2.42 <math><mo>%</mo></math> and 2.21 <math><mo>%</mo></math> for PH2 and ISIC2017, respectively, but also neural networks utilizing this loss function generally surpass those that do not in terms of segmentation performance.</p><p><strong>Conclusion: </strong>This work proposed a loss function that mined the information of specific pixels deeply without incurring additional training costs, significantly improving the automation of neural networks in image segmentation tasks. This loss function adapts to dermoscopic images of varying qualities and demonstrates higher effectiveness and robustness compared to other boundary loss functions, making it suitable for image segmentation tasks across various neural networks.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":null,"pages":null},"PeriodicalIF":2.3,"publicationDate":"2024-08-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141876608","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}
S F Frisken, N Haouchine, D D Chlorogiannis, V Gopalakrishnan, A Cafaro, W T Wells, A J Golby, R Du
{"title":"VESCL: an open source 2D vessel contouring library.","authors":"S F Frisken, N Haouchine, D D Chlorogiannis, V Gopalakrishnan, A Cafaro, W T Wells, A J Golby, R Du","doi":"10.1007/s11548-024-03212-0","DOIUrl":"10.1007/s11548-024-03212-0","url":null,"abstract":"<p><strong>Purpose: </strong>VESCL (pronounced 'vessel') is a novel vessel contouring library for computer-assisted 2D vessel contouring and segmentation. VESCL facilitates manual vessel segmentation in 2D medical images to generate gold-standard datasets for training, testing, and validating automatic vessel segmentation.</p><p><strong>Methods: </strong>VESCL is an open-source C++ library designed for easy integration into medical image processing systems. VESCL provides an intuitive interface for drawing variable-width parametric curves along vessels in 2D images. It includes highly optimized localized filtering to automatically fit drawn curves to the nearest vessel centerline and automatically determine the varying vessel width along each curve. To support a variety of segmentation paradigms, VESCL can export multiple segmentation representations including binary segmentations, occupancy maps, and distance fields.</p><p><strong>Results: </strong>VESCL provides sub-pixel resolution for vessel centerlines and vessel widths. It is optimized to segment small vessels with single- or sub-pixel widths that are visible to the human eye but hard to segment automatically via conventional filters. When tested on neurovascular digital subtraction angiography (DSA), VESCL's intuitive hand-drawn input with automatic curve fitting increased the speed of fully manual segmentation by 22× over conventional methods and by 3× over the best publicly available computer-assisted manual segmentation method. Accuracy was shown to be within the range of inter-operator variability of gold standard manually segmented data from a publicly available dataset of neurovascular DSA images as measured using Dice scores. Preliminary tests showed similar improvements for segmenting DSA of coronary arteries and RGB images of retinal arteries.</p><p><strong>Conclusion: </strong>VESCL is an open-source C++ library for contouring vessels in 2D images which can be used to reduce the tedious, labor-intensive process of manually generating gold-standard segmentations for training, testing, and comparing automatic segmentation methods.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":null,"pages":null},"PeriodicalIF":2.3,"publicationDate":"2024-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141328066","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}
Tejas Sudharshan Mathai, Bohan Liu, Ronald M Summers
{"title":"Segmentation of mediastinal lymph nodes in CT with anatomical priors.","authors":"Tejas Sudharshan Mathai, Bohan Liu, Ronald M Summers","doi":"10.1007/s11548-024-03165-4","DOIUrl":"10.1007/s11548-024-03165-4","url":null,"abstract":"<p><strong>Purpose: </strong>Lymph nodes (LNs) in the chest have a tendency to enlarge due to various pathologies, such as lung cancer or pneumonia. Clinicians routinely measure nodal size to monitor disease progression, confirm metastatic cancer, and assess treatment response. However, variations in their shapes and appearances make it cumbersome to identify LNs, which reside outside of most organs.</p><p><strong>Methods: </strong>We propose to segment LNs in the mediastinum by leveraging the anatomical priors of 28 different structures (e.g., lung, trachea etc.) generated by the public TotalSegmentator tool. The CT volumes from 89 patients available in the public NIH CT Lymph Node dataset were used to train three 3D off-the-shelf nnUNet models to segment LNs. The public St. Olavs dataset containing 15 patients (out-of-training-distribution) was used to evaluate the segmentation performance.</p><p><strong>Results: </strong>For LNs with short axis diameter <math><mo>≥</mo></math> 8 mm, the 3D cascade nnUNet model obtained the highest Dice score of 67.9 ± 23.4 and lowest Hausdorff distance error of 22.8 ± 20.2. For LNs of all sizes, the Dice score was 58.7 ± 21.3 and this represented a <math><mo>≥</mo></math> 10% improvement over a recently published approach evaluated on the same test dataset.</p><p><strong>Conclusion: </strong>To our knowledge, we are the first to harness 28 distinct anatomical priors to segment mediastinal LNs, and our work can be extended to other nodal zones in the body. The proposed method has the potential for improved patient outcomes through the identification of enlarged nodes in initial staging CT scans.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":null,"pages":null},"PeriodicalIF":2.3,"publicationDate":"2024-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11329534/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140916595","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}
Jessica Copeland, Mehida Rojas-Alexandre, Lilian Tsai, Franklin King, Nobuhiko Hata
{"title":"Characterizing the accuracy of robotic bronchoscopy in localization & targeting of small pulmonary lesions.","authors":"Jessica Copeland, Mehida Rojas-Alexandre, Lilian Tsai, Franklin King, Nobuhiko Hata","doi":"10.1007/s11548-024-03152-9","DOIUrl":"10.1007/s11548-024-03152-9","url":null,"abstract":"<p><strong>Purpose: </strong>Considering the recent implementation of lung cancer screening guidelines, it is crucial that small pulmonary nodules are accurately diagnosed. There is a significant need for quick, precise, and minimally invasive biopsy methods, especially for patients with small lung lesions in the outer periphery. Robotic bronchoscopy (RB) has recently emerged as a novel solution. The purpose of this study was to evaluate the accuracy of RB compared to the existing standard, electromagnetic navigational bronchoscopy (EM-NB).</p><p><strong>Methods: </strong>A prospective, single-blinded, and randomized-controlled study was performed to compare the accuracy of RB to EM-NB in localizing and targeting pulmonary lesions in a porcine lung model. Four operators were tasked with navigating to four pulmonary targets in the outer periphery of a porcine lung, to which they were blinded, using both the RB and EM-NB systems. The dependent variable was accuracy. Accuracy was measured as a rate of success in lesion localization and targeting, the distance from the center of the pulmonary target, and by anatomic location. The independent variable was the navigation system, RB was compared to EM-NB using 1:1 randomization.</p><p><strong>Results: </strong>Of 75 attempts, 72 were successful in lesion localization and 60 were successful in lesion targeting. The success rate for lesion localization was 100% with RB and 91% with EM- NB. The success rate for lesion targeting was 93% with RB and 80% for EM-NB. RB demonstrated superior accuracy in reaching the distance from the center of the lesion, at 0.62 mm compared to EM-NB at 1.28 mm (p = 0.001). Accuracy was improved using RB compared to EM- NB for lesions in the LLL (p = 0.025), LUL (p < 0.001), and RUL (p < 0.001).</p><p><strong>Conclusion: </strong>Our findings support RB as a more accurate method of navigating and localizing small peripheral pulmonary targets when compared to standard EM-NB in a porcine lung model. This may be attributed to the ability of RB to reduce substantial tissue displacement seen with standard EM-NB navigation. As the development and application of RB advances, so will the ability to accurately diagnose small peripheral lung cancer nodules, providing patients with early-stage lung cancer the best possible outcomes.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":null,"pages":null},"PeriodicalIF":2.3,"publicationDate":"2024-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141421761","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}
Ina Vernikouskaya, Hans-Peter Müller, Albert C Ludolph, Jan Kassubek, Volker Rasche
{"title":"AI-assisted automatic MRI-based tongue volume evaluation in motor neuron disease (MND).","authors":"Ina Vernikouskaya, Hans-Peter Müller, Albert C Ludolph, Jan Kassubek, Volker Rasche","doi":"10.1007/s11548-024-03099-x","DOIUrl":"10.1007/s11548-024-03099-x","url":null,"abstract":"<p><strong>Purpose: </strong>Motor neuron disease (MND) causes damage to the upper and lower motor neurons including the motor cranial nerves, the latter resulting in bulbar involvement with atrophy of the tongue muscle. To measure tongue atrophy, an operator independent automatic segmentation of the tongue is crucial. The aim of this study was to apply convolutional neural network (CNN) to MRI data in order to determine the volume of the tongue.</p><p><strong>Methods: </strong>A single triplanar CNN of U-Net architecture trained on axial, coronal, and sagittal planes was used for the segmentation of the tongue in MRI scans of the head. The 3D volumes were processed slice-wise across the three orientations and the predictions were merged using different voting strategies. This approach was developed using MRI datasets from 20 patients with 'classical' spinal amyotrophic lateral sclerosis (ALS) and 20 healthy controls and, in a pilot study, applied to the tongue volume quantification to 19 controls and 19 ALS patients with the variant progressive bulbar palsy (PBP).</p><p><strong>Results: </strong>Consensus models with softmax averaging and majority voting achieved highest segmentation accuracy and outperformed predictions on single orientations and consensus models with union and unanimous voting. At the group level, reduction in tongue volume was not observed in classical spinal ALS, but was significant in the PBP group, as compared to controls.</p><p><strong>Conclusion: </strong>Utilizing single U-Net trained on three orthogonal orientations with consequent merging of respective orientations in an optimized consensus model reduces the number of erroneous detections and improves the segmentation of the tongue. The CNN-based automatic segmentation allows for accurate quantification of the tongue volumes in all subjects. The application to the ALS variant PBP showed significant reduction of the tongue volume in these patients and opens the way for unbiased future longitudinal studies in diseases affecting tongue volume.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":null,"pages":null},"PeriodicalIF":2.3,"publicationDate":"2024-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11329588/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140307730","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":"Using diffusion models to generate synthetic labeled data for medical image segmentation.","authors":"Daniel G Saragih, Atsuhiro Hibi, Pascal N Tyrrell","doi":"10.1007/s11548-024-03213-z","DOIUrl":"10.1007/s11548-024-03213-z","url":null,"abstract":"<p><strong>Purpose: </strong>Medical image analysis has become a prominent area where machine learning has been applied. However, high-quality, publicly available data are limited either due to patient privacy laws or the time and cost required for experts to annotate images. In this retrospective study, we designed and evaluated a pipeline to generate synthetic labeled polyp images for augmenting medical image segmentation models with the aim of reducing this data scarcity.</p><p><strong>Methods: </strong>We trained diffusion models on the HyperKvasir dataset, comprising 1000 images of polyps in the human GI tract from 2008 to 2016. Qualitative expert review, Fréchet Inception Distance (FID), and Multi-Scale Structural Similarity (MS-SSIM) were tested for evaluation. Additionally, various segmentation models were trained with the generated data and evaluated using Dice score (DS) and Intersection over Union (IoU).</p><p><strong>Results: </strong>Our pipeline produced images more akin to real polyp images based on FID scores. Segmentation model performance also showed improvements over GAN methods when trained entirely, or partially, with synthetic data, despite requiring less compute for training. Moreover, the improvement persists when tested on different datasets, showcasing the transferability of the generated images.</p><p><strong>Conclusions: </strong>The proposed pipeline produced realistic image and mask pairs which could reduce the need for manual data annotation when performing a machine learning task. We support this use case by showing that the methods proposed in this study enhanced segmentation model performance, as measured by Dice and IoU scores, when trained fully or partially on synthetic data.</p>","PeriodicalId":51251,"journal":{"name":"International Journal of Computer Assisted Radiology and Surgery","volume":null,"pages":null},"PeriodicalIF":2.3,"publicationDate":"2024-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141428268","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}