Surgical Video Research Articles

DBH-YOLO: a surgical instrument detection method based on feature separation in laparoscopic surgery.

Accurately locating and analysing surgical instruments in laparoscopic surgical videos can assist doctors in postoperative quality assessment. This can provide patients with more scientific and rational solutions for healing surgical complications. Therefore, we propose an end-to-end algorithm for the detection of surgical instruments. Dual-Branched Head (DBH) and Overall Intersection over Union Loss (OIoU Loss) are introduced to solve the problem of inaccurate surgical instrument detection, both in terms of localization and classification. An effective method (DBHYOLO) for the detection for laparoscopic surgery in complex scenarios is proposed. This study manually annotates a new laparoscopic gastric cancer resection surgical instrument location dataset LGIL, which provides a better validation platform for surgical instrument detection methods. The proposed method's performance was tested using the m2cai16-tool-locations, LGIL, and Onyeogulu datasets. The mean Average Precision (mAP) values obtained were 96.8%, 95.6%, and 98.4%, respectively, which were higher than the other classical models compared. The improved model is more effective than the benchmark network in distinguishing between surgical instrument classes with high similarity and avoiding too many missed detection cases. In this paper, the problem of inaccurate detection of surgical instruments is addressed from two different perspectives: classification and localization. And the experimental results on three representative datasets verify the performance of DBH-YOLO. It is shown that this method has a good generalization capability.

Cataract-1K Dataset for Deep-Learning-Assisted Analysis of Cataract Surgery Videos

In recent years, the landscape of computer-assisted interventions and post-operative surgical video analysis has been dramatically reshaped by deep-learning techniques, resulting in significant advancements in surgeons’ skills, operation room management, and overall surgical outcomes. However, the progression of deep-learning-powered surgical technologies is profoundly reliant on large-scale datasets and annotations. In particular, surgical scene understanding and phase recognition stand as pivotal pillars within the realm of computer-assisted surgery and post-operative assessment of cataract surgery videos. In this context, we present the largest cataract surgery video dataset that addresses diverse requisites for constructing computerized surgical workflow analysis and detecting post-operative irregularities in cataract surgery. We validate the quality of annotations by benchmarking the performance of several state-of-the-art neural network architectures for phase recognition and surgical scene segmentation. Besides, we initiate the research on domain adaptation for instrument segmentation in cataract surgery by evaluating cross-domain instrument segmentation performance in cataract surgery videos. The dataset and annotations are publicly available in Synapse.

Nonpigmented Adenoma of the Retinal Pigment Epithelium Mimicking a Retinal Vasoproliferative Tumor Managed With Endoresection

Purpose: To present a case with signs suggestive of a retinal vasoproliferative tumor. Methods: A case report was evaluated and a surgical video presented. Results: A 61-year-old White man presented with an amelanotic retinal tumor associated with exudation, retinal edema, and overlying telangiectatic vessels, suggestive of a retinal vasoproliferative tumor. Standardized echography showed an irregular mass with medium-to-high internal reflectivity and internal calcification, which suggested chronicity. He was initially treated for an exudative retinal detachment (RD) in the context of a presumed vasoproliferative tumor but later developed combined exudative and rhegmatogenous RD, prompting surgical repair with tumor endoresection. Pathology showed nonpigmented adenoma of the retinal pigment epithelium (RPE). Conclusions: Nonpigmented adenoma of the RPE is a rare tumor, and its clinical similarity to a vasoproliferative tumor should be noted. Endoresection may be considered in cases resulting in RD.

Perfluorocarbon Liquid-Silicone Oil Exchange: The Simplicity and Reliability of the Single Port Technique and a Review of Literature

Purpose: To report the “Single Port Technique” for perfluorocarbon liquid (PFCL)-silicone oil (SO) exchange for the management of giant retinal tear detachments (GRTD). Methods: The previously reported techniques of direct PFCL-SO exchange used 2 ports to achieve the influx and outflux of fluid. The term 'Single Port' refers to the use of only one port as the exclusive port for the inflow of SO and the outflow of fluid meniscus and PFCL, performed alternately. This technique is proposed, with illustrated instructions and a surgical video. Results: The newly proposed “Single Port Technique” offers a straightforward step using a switch between active aspiration by the vitreous cutter and active silicone oil injection through a single port without the need for assistance. This method has been performed by the authors on 29 patients, including 22 GRTD patients and 7 patients with rhegmatogenous retinal detachment caused by dialysis, and it has demonstrated its simplicity and reliability for use with only standard vitrectomy tools and machine settings. Conclusion: The Single Port Technique for PFCL-SO exchange for the management of GRTD and rhegmatogenous retinal detachment caused by dialysis has been tried and tested, and it is a valuable tool that can be easily used by any vitreoretinal surgeon.

Holmium Laser Incision of Ejaculatory Duct (HoLIED): Surgical Procedure

Introduction: Ejaculatory duct obstruction is a well-known cause of infertility and many other symptoms, such as hematospermia and lower urinary tract symptoms. Diagnosis is made based on the medical history and a series of imaging tests identifying the obstructed ducts. Transurethral resection of the ejaculatory ducts is the mainstay treatment to release the obstruction. In this step-by-step surgical video, we will describe a novel technique using the holmium laser to incise the ejaculatory ducts in a patient with chronic lower urinary tract symptoms, hematospermia, and persistent urethral discharge. Materials and Methods: A pelvic MRI was obtained in workup of the above issues and identified a complex ejaculatory duct cyst. The patient consented for Holmium Laser Incision of Ejaculatory Duct. A holmium laser fiber was used through a 24F rigid cystoscope to unroof the ejaculatory duct cyst, resulting in the drainage of a dark brown fluid. Following unroofing of the cyst, the ejaculatory ducts were identified using a 17F rigid cystoscope and a 16F Foley catheter was inserted. Results: The Foley catheter was removed on POD1, and the patient was able to void. His symptoms improved progressively with time and antibiotics. A repeat MRI was obtained several months postoperatively and showed the resolution of the cystic lesion. Conclusion: Holmium Laser Incision of Ejaculatory Duct is safe and feasible procedure allowing to treat ejaculatory duct obstruction without increased risk of complications.

Attitudes and access to surgical video recording among ophthalmology trainees in Scotland.

Attitudes and access to surgical video recording among ophthalmology trainees in Scotland.

Assessment of the Quality of YouTube Educational Videos on Laparoscopic Right Adrenalectomy and Laparoscopic Partial Nephrectomy Surgeries.

Introduction: Despite the considerable studies conducted on the quality of YouTube surgical videos as an educational resource in other surgical fields, there have been no such studies in the field of laparoscopic urology so far. Considering the great sensitivity in these procedures and the necessity of identifying the mistakes in these videos, we aimed to evaluate the quality of YouTube educational videos on laparoscopic right adrenalectomy and laparoscopic partial nephrectomy surgeries. Materials and Methods: In this descriptive cross-sectional study, 131 YouTube educational videos on laparoscopic right adrenalectomy and laparoscopic partial nephrectomy surgeries were reviewed. Two researchers familiar with laparoscopic urological surgery reviewed the videos based on the LAP-VEGaS checklist. A third professor reviewed the videos on which there were disagreements. Results: In the majority of the videos, the title was chosen accordingly and the surgeon was introduced appropriately. Furthermore, in most of the videos, patient anonymity was respected. The mean score of the videos was equal to 74.3 ± 5.4, the maximum score being 17 and the minimum 1. The average score of the partial nephrectomy videos was obtained as 98.3 ± 5.74, whereas the average score for adrenalectomy videos was 47.3 ± 4.5. The overall average score of the studied videos was 74.3 ± 5.40. Conclusion: The results of our study suggest that most of YouTube videos on laparoscopic right adrenalectomy and laparoscopic partial nephrectomy surgeries are used for educational purpose. These YouTube videos are suboptimal in educational aspect and students should be advised to use them with caution.

MT4MTL-KD: A Multi-Teacher Knowledge Distillation Framework for Triplet Recognition.

The recognition of surgical triplets plays a critical role in the practical application of surgical videos. It involves the sub-tasks of recognizing instruments, verbs, and targets, while establishing precise associations between them. Existing methods face two significant challenges in triplet recognition: 1) the imbalanced class distribution of surgical triplets may lead to spurious task association learning, and 2) the feature extractors cannot reconcile local and global context modeling. To overcome these challenges, this paper presents a novel multi-teacher knowledge distillation framework for multi-task triplet learning, known as MT4MTL-KD. MT4MTL-KD leverages teacher models trained on less imbalanced sub-tasks to assist multi-task student learning for triplet recognition. Moreover, we adopt different categories of backbones for the teacher and student models, facilitating the integration of local and global context modeling. To further align the semantic knowledge between the triplet task and its sub-tasks, we propose a novel feature attention module (FAM). This module utilizes attention mechanisms to assign multi-task features to specific sub-tasks. We evaluate the performance of MT4MTL-KD on both the 5-fold cross-validation and the CholecTriplet challenge splits of the CholecT45 dataset. The experimental results consistently demonstrate the superiority of our framework over state-of-the-art methods, achieving significant improvements of up to 6.4% on the cross-validation split.

Introducing surgical intelligence in gynecology: Automated identification of key steps in hysterectomy.

The analysis of surgical videos using artificial intelligence holds great promise for the future of surgery by facilitating the development of surgical best practices, identifying key pitfalls, enhancing situational awareness, and disseminating that information via real-time, intraoperative decision-making. The objective of the present study was to examine the feasibility and accuracy of a novel computer vision algorithm for hysterectomy surgical step identification. This was a retrospective study conducted on surgical videos of laparoscopic hysterectomies performed in 277 patients in five medical centers. We used a surgical intelligence platform (Theator Inc.) that employs advanced computer vision and AI technology to automatically capture video data during surgery, deidentify, and upload procedures to a secure cloud infrastructure. Videos were manually annotated with sequential steps of surgery by a team of annotation specialists. Subsequently, a computer vision system was trained to perform automated step detection in hysterectomy. Analyzing automated video annotations in comparison to manual human annotations was used to determine accuracy. The mean duration of the videos was 103 ± 43 min. Accuracy between AI-based predictions and manual human annotations was 93.1% on average. Accuracy was highest for the dissection and mobilization step (96.9%) and lowest for the adhesiolysis step (70.3%). The results of the present study demonstrate that a novel AI-based model achieves high accuracy for automated steps identification in hysterectomy. This lays the foundations for the next phase of AI, focused on real-time clinical decision support and prediction of outcome measures, to optimize surgeon workflow and elevate patient care.

Assessing the educational value of laparoscopic radical nephrectomy videos on YouTube®: A comparative analysis of short versus long videos.

To evaluate the quality of laparoscopic radical nephrectomy videos and determine the extent to which they are informative and educational for healthcare professionals. We used the YouTube® search engine to search for the term 'laparoscopic radical nephrectomy' with time filters of 4-20 min (Group 1) and >20 min (Group 2) and then sorted the results uploaded chronologically before January 2023. One hundred videos were analysed for each group. The reliability of the videos was assessed using the Journal of American Medical Association (JAMA) Benchmark Criteria and DISCERN questionnaire scores (DISCERN). Educational quality was assessed using the Global Quality Score (GQS) and a 20-item objective scoring system (OSS) for laparoscopic nephrectomy. The popularity of the videos was evaluated using the video power index (VPI). The mean video duration was 8.9 ± 4.3 min in Group 1 and 52.02 ± 31.09 min in Group 2 (P < 0.001). The mean JAMA (2.49 ± 0.61) and OSS scores (60 ± 12.3) were higher in Group 2 than in Group 1, while no significant difference was observed in the mean GQS (2.53 ± 0.7, 2.39 ± 0.88, respectively) between the groups (P < 0.001, P = 0.039, P = 0.131, respectively). While the standardisation of surgical videos published on YouTube® and the establishment of auditing mechanisms do not seem plausible, high total OSS, periprocedural OSS, and VPI scores, and high OSS, JAMAS, GQS and DISCERN scores in long videos indicate that such videos offer a greater contribution to education.

Privacy-Preserving Algorithms Can Facilitate Surgical Video Databases for Quality Improvement, Education, and Research.

Privacy-Preserving Algorithms Can Facilitate Surgical Video Databases for Quality Improvement, Education, and Research.

Assessing anatomical variations of the inferior mesenteric artery via three-dimensional CT angiography and laparoscopic colorectal surgery: a retrospective observational study

To assess the anatomy of the inferior mesenteric artery (IMA) and its branches by reviewing laparoscopic left-sided colorectal cancer surgery videos and comparing them with preoperative three-dimensional computed tomography (3D-CT) angiography, to verify the accuracy of 3D-CT vascular reconstruction techniques. High-definition surgical videos and preoperative imaging data of 200 patients who underwent laparoscopic left-sided colorectal cancer surgery were analysed, and the alignment of the IMA and its branches in relation to the inferior mesenteric vein (IMV) was observed and summarized. The above two methods were used to measure the length of the IMA and its branches. Of 200 patients, 47.0% had the sigmoid arteries (SAs) arise from the common trunk with the superior rectal artery (SRA), and 30.5% had the SAs arise from the common trunk with the left colic artery (LCA). In 3.5% of patients, the SAs arising from both the LCA and SRA. The LCA, SA, and SRA emanated from the same point in 13.5% of patients, and the LCA was absent in 5.5% of patients. The range of D cm (IMA length measured by intraoperative silk thread) and d cm (IMA length measured by 3D-CT vascular reconstruction) in all cases was 1.84–6.62 cm and 1.85–6.52 cm, respectively, and there was a significant difference between them. (p < 0.001). The lengths between the intersection of the LCA and IMV measured intraoperatively were 0.64–4.29 cm, 0.87–4.35 cm, 1.32–4.28 cm and 1.65–3.69 cm in types 1A, 1B, 1C, and 2, respectively, and there was no significant difference between the groups (p = 0.994). There was only a significant difference in the length of the IMA between the 3D-CT vascular reconstruction and intraoperative observation data, which can provide guidance to surgeons in preoperative preparation.

Assessment of Surgical Training Modality Efficacy in Medical Students – A Comparison of Surgical Simulation, Video Demonstration, and Live Instruction

Objectives Surgical simulation is a growing form of education in medical training, but it remains unclear how simulation compares to traditional lecture or apprenticeship models of teaching when translating procedural knowledge to surgical performance. This study aimed to assess the effectiveness of translating procedural knowledge to surgical skill when trained by the surgical simulation app “Touch Surgery (TS),” recorded video lecture, or in-person teaching of the trigger finger release (TFR) surgery. The study also aimed to identify the learners’ preferred teaching modality. Methods Twenty-seven first- and second-year medical students were recruited to participate. They were randomly assigned to one of three cohorts (n=9) as designated by a different TFR surgery learning modality: surgical simulation through TS (cohort “SS”), video demonstration (cohort “VIDEO”), or live teaching by a board-certified orthopaedic hand surgeon (cohort “LIVE”). Each participant completed their modality three times. They then completed an assessment of the procedural steps before performing the TFR surgery on a cadaver. Outcome measures included procedural test scores and cadaveric performance, evaluated using a surgery-specific detailed checklist, a global rating scale (GRS) of soft surgical skills, and a pass/fail assessment. Participants graded their modality’s usefulness using a 5-point Likert scale. Results There was no statistically significant difference in TFR procedural assessment scores (P=0.123) or cadaveric surgical performance between groups when evaluated by the surgery-specific step checklist (P=0.549), GRS (P=0.567), and pass/fail assessment (P=0.874). Students in the LIVE cohort rated their modality as their first-choice training tool (P=0.009); however, those in the SS cohort rated their modality as easiest to use, follow, and understand (P=0.010). Conclusion All educational modalities should be considered in surgical training. This study demonstrates that students can perform cognitively and technically similar learning from virtual or live formats; however, they preferred live teaching. Level of Evidence III

A Multimodal Transformer Model for Recognition of Images from Complex Laparoscopic Surgical Videos.

The determination of the potential role and advantages of artificial intelligence-based models in the field of surgery remains uncertain. This research marks an initial stride towards creating a multimodal model, inspired by the Video-Audio-Text Transformer, that aims to reduce negative occurrences and enhance patient safety. The model employs text and image embedding state-of-the-art models (ViT and BERT) to assess their efficacy in extracting the hidden and distinct features from the surgery video frames. These features are then used as inputs for convolution-free Transformer architectures to extract comprehensive multidimensional representations. A joint space is then used to combine the text and image features extracted from both Transformer encoders. This joint space ensures that the relationships between the different modalities are preserved during the combination process. The entire model was trained and tested on laparoscopic cholecystectomy (LC) videos encompassing various levels of complexity. Experimentally, a mean accuracy of 91.0%, a precision of 81%, and a recall of 83% were reached by the model when tested on 30 videos out of 80 from the Cholec80 dataset.

Selection of a Surgical Approach for Middle Ear Cholesteatoma Based on the Fusion Images of Non-Echo Planar Diffusion-Weighted MRI and CT.

This study aimed to explore the ability of fusion images of non-echo planar diffusion-weighted magnetic resonance imaging (non-EPI-DWI MRI) and computed tomography (CT) to accurately locate cholesteatoma and plan the surgical approach. In the first part, 41 patients were included. Their CT images and non-EPI DWMRI images were fused. The scope of cholesteatoma in the fusion image was compared with that in the surgical video to evaluate the capability to locate cholesteatoma. A total of 229 patients were included in the second part, and they were divided into 2 groups. We chose the surgical approach for the CT group and the fusion group, and compared the accuracy of surgical approaches in the CT group and the fusion group using the surgical records. The location of cholesteatoma shown in the fusion images was almost identical to that observed during the operation (kappa = .862). The overall specificity and sensitivity of the fusion images in locating cholesteatoma were 94.12% and 93.06%, respectively. The accuracy of surgical approach selection based on the fusion images (99.02%) was higher than that of surgical approach selection based on the CT images (85.83%). It is recommended that the fusion images be used to locate the range of the cholesteatoma before operation.

Microsurgical Resection of a Medulla Oblongata Cavernoma: 3-Dimensional Operative Video.

Cavernous malformations occur in approximately 0.5% of the population. When it comes to the medulla oblongata, the incidence rate is around 5%.1,2 Patients with these conditions face potential risks such as hemorrhaging, substantial mortality, and morbidity. The symptoms experienced by the patient vary depending on the extent of hemorrhaging and the location of the lesion. Surgery stands as the primary and essential form of treatment in these cases.1-4 The aim of this operative video was to present the removal of a medulla oblongata cavernoma. The patient consented to the procedure and to the publication of her images and surgical video. We describe the case of a 53-year-old woman presenting with headaches and with swallowing difficulties, being tracheostomized on hospital admission. Magnetic resonance imaging showed a lesion located at the medulla oblongata. Owing to the patient's symptoms and cavernoma bleeding, surgical resection was determined. There were no adverse outcomes after surgery, and the patient evolved favorably. Postoperative imaging displayed complete removal of the cavernoma. Microsurgical resection of cavernomas in this location with neurophysiological monitoring and selection of the safe entry zone is an effective option.1-4.

The development of a deep learning model for automated segmentation of the robotic pancreaticojejunostomy.

Minimally invasive surgery provides an unprecedented opportunity to review video for assessing surgical performance. Surgical video analysis is time-consuming and expensive. Deep learning provides an alternative for analysis. Robotic pancreaticoduodenectomy (RPD) is a complex and morbid operation. Surgeon technical performance of pancreaticojejunostomy (PJ) has been associated with postoperative pancreatic fistula. In this work, we aimed to utilize deep learning to automatically segment PJ RPD videos. This was a retrospective review of prospectively collected videos from 2011 to 2022 that were in libraries at tertiary referral centers, including 111 PJ videos. Each frame of a robotic PJ video was categorized based on 6 tasks. A 3D convolutional neural network was trained for frame-level visual feature extraction and classification. All the videos were manually annotated for the start and end of each task. Of the 100 videos assessed, 60 videos were used for the training the model, 10 for hyperparameter optimization, and 30 for the testing of performance. All the frames were extracted (6 frames/second) and annotated. The accuracy and mean per-class F1 scores were 88.01% and 85.34% for tasks. The deep learning model performed well for automated segmentation of PJ videos. Future work will focus on skills assessment and outcome prediction.

Evaluation of the approach of General Surgery Specialists in Turkey to surgical videos published on YouTube

Objective: YouTube is an extremely popular social media platform used to watch, upload and share videos. YouTube contains many surgical videos intended for educational or marketing purposes. This article primarily aims to reveal Turkish general surgeons’ approach to and interest in social media, with a special focus on YouTube, and to discuss the pros and cons of YouTube as an educational source.&#x0D; Methods: An online survey of general surgery residents and attending physicians was conducted between April 2021 and May 2021 using a 16-item questionnaire. Participants were first asked to provide consent and provide demographic data such as age, sex and medical title, and then to answer questions about the use of social media, particularly YouTube, viewing surgical videos on YouTube, the ethical aspect of posting surgical videos on social media and whether or not they use it for educational purposes. Statistical analysis was conducted using descriptive statistics, ratios and frequencies.&#x0D; Results: The survey was completed by 46 general surgeons with various medical titles from different healthcare institutions. Most participants reported using YouTube to watch videos (95.7%). Almost a quarter (28.3%) reported frequently using YouTube to view surgery videos. Furthermore, most participants (87%) found the surgery videos on YouTube educational (91.3%). One of the striking results was that 52.2% of the participants reported that they had a friend who performed a first-time surgical procedure after watching it on YouTube. In addition, most surgeons considered it ethically appropriate to use surgical videos from YouTube as an educational source (76.10%).&#x0D; Conclusion: YouTube draws great interest from general surgeons as well as other members of society. General surgeons watch surgical videos on YouTube for educational purposes and find it ethical to post such videos. However, more extensive studies are needed to clarify further the role of increasingly used video-sharing platforms in education.

Current trends and barriers to video management and analytics as a tool for surgeon skilling.

The benefits of intraoperative recording are well published in the literature; however, few studies have identified current practices, barriers, and subsequent solutions. The objective of this study was to better understand surgeon's current practices and perceptions of video management and gather blinded feedback on a new surgical video recording product with the potential to address these barriers effectively. A structured questionnaire was used to survey 230 surgeons (general, gynecologic, and urologic) and hospital administrators across the US and Europe regarding their current video recording practices. The same questionnaire was used to evaluate a blinded concept describing a new intraoperative recording solution. 54% of respondents reported recording eligible cases, with the majority recording less than 35% of their total eligible caseload. Reasons for not recording included finding no value in recording simple procedures, forgetting to record, lack of access to equipment, legal concerns, labor intensity, and difficulty accessing videos. Among non-recording surgeons, 65% reported considering recording cases to assess surgical techniques, document practice, submit to conferences, share with colleagues, and aid in training. 35% of surgeons rejected recording due to medico-legal concerns, lack of perceived benefit, concerns about secure storage, and price. Regarding the concept of a recording solution, 74% of all respondents were very likely or quite likely to recommend the product for adoption at their facility. Appealing features to current recorders included the product's ease of use, use of AI to maintain patient and staff privacy, lack of manual downloads, availability of full-length procedural videos, and ease of access and storage. Non-recorders found the immediate access to videos and maintenance of patient/staff privacy appealing. Tools that address barriers to recording, accessing, and managing surgical case videos are critical for improving surgical skills. Touch Surgery Enterprise is a valuable tool that can help overcome these barriers.

Real-Time Ultrasound-Guided Posterolateral Thoracic Discectomy: 2-Dimensional Operative Video.

Symptomatic thoracic disk herniations can be challenging to treat. Various surgical approaches have been reported in the literature, but the ideal approach is still a topic of controversy. We present a step-by-step surgical video highlighting the utilization of continuous, real-time intraoperative ultrasound (IOUS) guidance for thoracic discectomy through a posterolateral approach. The background and rationale for the procedure and a detailed presentation of the sequence of surgical steps are detailed in the operative video. We believe that real-time IOUS can be used not only to gain an understanding of the ventral anatomy and to verify decompression after discectomy but it can also be used in a real-time fashion to guide the maneuvers for discectomy itself. Advantages of using real-time IOUS include real-time monitoring of the spinal cord during mobilization of the disk-osteophyte complex and direct visualization of the dura-disk herniation interface, which may enhance the safety of surgery. The patient consented to the procedure and the publication of her image.