Dataset Information

Qualitative Comparison of Image Stitching Algorithms for Multi-Camera Systems in Laparoscopy.

ABSTRACT: Multi-camera systems were recently introduced into laparoscopy to increase the narrow field of view of the surgeon. The video streams are stitched together to create a panorama that is easier for the surgeon to comprehend. Multi-camera prototypes for laparoscopy use quite basic algorithms and have only been evaluated on simple laparoscopic scenarios. The more recent state-of-the-art algorithms, mainly designed for the smartphone industry, have not yet been evaluated in laparoscopic conditions. We developed a simulated environment to generate a dataset of multi-view images displaying a wide range of laparoscopic situations, which is adaptable to any multi-camera system. We evaluated classical and state-of-the-art image stitching techniques used in non-medical applications on this dataset, including one unsupervised deep learning approach. We show that classical techniques that use global homography fail to provide a clinically satisfactory rendering and that even the most recent techniques, despite providing high quality panorama images in non-medical situations, may suffer from poor alignment or severe distortions in simulated laparoscopic scenarios. We highlight the main advantages and flaws of each algorithm within a laparoscopic context, identify the main remaining challenges that are specific to laparoscopy, and propose methods to improve these approaches. We provide public access to the simulated environment and dataset.

SUBMITTER: Guy S

PROVIDER: S-EPMC8951246 | biostudies-literature | 2022 Feb

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Qualitative Comparison of Image Stitching Algorithms for Multi-Camera Systems in Laparoscopy.

Guy Sylvain S Haberbusch Jean-Loup JL Promayon Emmanuel E Mancini Stéphane S Voros Sandrine S

Journal of imaging 20220223 3

Multi-camera systems were recently introduced into laparoscopy to increase the narrow field of view of the surgeon. The video streams are stitched together to create a panorama that is easier for the surgeon to comprehend. Multi-camera prototypes for laparoscopy use quite basic algorithms and have only been evaluated on simple laparoscopic scenarios. The more recent state-of-the-art algorithms, mainly designed for the smartphone industry, have not yet been evaluated in laparoscopic conditions. W ...[more]

PMID: 35324607

Similar Datasets

Project description:BackgroundInexperienced operating assistants are often tasked with the important role of handling camera navigation during laparoscopic surgery. Incorrect handling can lead to poor visualization, increased operating time, and frustration for the operating surgeon-all of which can compromise patient safety. The objectives of this trial were to examine how to train laparoscopic camera navigation and to explore the transfer of skills to the operating room.Materials and methodsA randomized, single-center superiority trial with three groups: The first group practiced simulation-based camera navigation tasks (camera group), the second group practiced performing a simulation-based cholecystectomy (procedure group), and the third group received no training (control group). Participants were surgical novices without prior laparoscopic experience. The primary outcome was assessment of camera navigation skills during a laparoscopic cholecystectomy. The secondary outcome was technical skills after training, using a previously developed model for testing camera navigational skills. The exploratory outcome measured participants' motivation toward the task as an operating assistant.ResultsThirty-six participants were randomized. No significant difference was found in the primary outcome between the three groups (p = 0.279). The secondary outcome showed no significant difference between the interventions groups, total time 167 s (95% CI, 118-217) and 194 s (95% CI, 152-236) for the camera group and the procedure group, respectively (p = 0.369). Both interventions groups were significantly faster than the control group, 307 s (95% CI, 202-412), p = 0.018 and p = 0.045, respectively. On the exploratory outcome, the control group for two dimensions, interest/enjoyment (p = 0.030) and perceived choice (p = 0.033), had a higher score.ConclusionsSimulation-based training improves the technical skills required for camera navigation, regardless of practicing camera navigation or the procedure itself. Transfer to the clinical setting could, however, not be demonstrated. The control group demonstrated higher interest/enjoyment and perceived choice than the camera group.

Project description:IntroductionSimulation training programs are essential for novice surgeons to acquire basic experience to master laparoscopic skills. However, current state-of-the-art laparoscopy simulators are still expensive, limiting the accessibility to practical training lessons. Furthermore, training is time intensive and requires extensive spatial capacity, limiting its availability to surgeons. New laparoscopic simulators offer a cost-effective alternative, which can be used to train in a digital environment, allowing flexible, digital and personalized laparoscopic training. This study investigates if training on low-cost simulators in a digital environment is comparable to in-person training formats.Materials and methodsFrom June 2023 to December 2023, 40 laparoscopic novices participated in this multi-center, prospective randomized controlled trial. All participants were randomized to either the ‟distance" (intervention) or the "in-person" (control) group. They were trained in a standardized laparoscopic training curriculum to reach a predefined level of proficiency. After completing the curriculum, participants performed four different laparoscopic tasks on the ForceSense system. Primary endpoints were overall task errors, the overall time for completion of the tasks, and force parameters.ResultsIn total, 40 laparoscopic novices completed digital or in-person training. Digital training showed no significant differences in developing basic laparoscopic skills compared to in-person training. There were no significant differences in median overall errors between both training groups for all exercises combined (intervention 3 vs. control 4; p value = 0.74). In contrast, the overall task completion time was significantly lower for the group trained digitally (intervention 827.92 s vs. control 993.42; p value = 0.015). The applied forces during the final assessment showed no significant differences between both groups for all exercises. Overall, over 90% of the participants rated the training as good or very good.ConclusionOur study shows that students that underwent digital laparoscopic training completed tasks with a similar number of errors but in a shorter time than students that underwent in-person training. Nevertheless, the best strategies to implement such digital training options need to be evaluated further to support surgeons' personal preferences and expectations.

Dataset Information

Qualitative Comparison of Image Stitching Algorithms for Multi-Camera Systems in Laparoscopy.

Publications

Qualitative Comparison of Image Stitching Algorithms for Multi-Camera Systems in Laparoscopy.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets