Dataset Information

Deep learning-based automatic surgical step recognition in intraoperative videos for transanal total mesorectal excision.

ABSTRACT:

Background

Dividing a surgical procedure into a sequence of identifiable and meaningful steps facilitates intraoperative video data acquisition and storage. These efforts are especially valuable for technically challenging procedures that require intraoperative video analysis, such as transanal total mesorectal excision (TaTME); however, manual video indexing is time-consuming. Thus, in this study, we constructed an annotated video dataset for TaTME with surgical step information and evaluated the performance of a deep learning model in recognizing the surgical steps in TaTME.

Methods

This was a single-institutional retrospective feasibility study. All TaTME intraoperative videos were divided into frames. Each frame was manually annotated as one of the following major steps: (1) purse-string closure; (2) full thickness transection of the rectal wall; (3) down-to-up dissection; (4) dissection after rendezvous; and (5) purse-string suture for stapled anastomosis. Steps 3 and 4 were each further classified into four sub-steps, specifically, for dissection of the anterior, posterior, right, and left planes. A convolutional neural network-based deep learning model, Xception, was utilized for the surgical step classification task.

Results

Our dataset containing 50 TaTME videos was randomly divided into two subsets for training and testing with 40 and 10 videos, respectively. The overall accuracy obtained for all classification steps was 93.2%. By contrast, when sub-step classification was included in the performance analysis, a mean accuracy (± standard deviation) of 78% (± 5%), with a maximum accuracy of 85%, was obtained.

Conclusions

To the best of our knowledge, this is the first study based on automatic surgical step classification for TaTME. Our deep learning model self-learned and recognized the classification steps in TaTME videos with high accuracy after training. Thus, our model can be applied to a system for intraoperative guidance or for postoperative video indexing and analysis in TaTME procedures.

SUBMITTER: Kitaguchi D

PROVIDER: S-EPMC8758657 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:ImportanceProponents of novel transanal total mesorectal excision (TME) suggest the procedure overcomes the technical and oncologic challenges of conventional approaches for treating rectal cancer. Recently, however, there has been controversy regarding the oncologic safety of the procedure.ObjectiveTo assess the association of transanal TME with the incidence of local recurrence (LR) of cancer and the probability of remaining free of LR during follow-up.Design, setting, and participantsThis multicenter cohort study used data from 8 high-volume rectal cancer academic institutions from across Canada on all consecutive patients with primary rectal cancer treated by transanal TME at the participating centers. The study was conducted between January 2014 and December 2018, and data were analyzed from April 1, 2020, to September 15, 2020.ExposureTransanal TME.Main outcomes and measuresThe incidence of LR was reported as a direct measure of quality of resection. The cumulative probability of LR- and systemic recurrence (SR)-free survival at 36 months was estimated. Local recurrence and SR were defined as radiologic or endoscopic evidence of 1 or more new lesions in or outside the pelvis, respectively, documented during surveillance after the removal of the primary tumor.ResultsOf 608 total patients included in the analysis, 423 (69.6%) were male; the median age was 63 years (interquartile range [IQR], 54-70 years). Local recurrence was identified in 22 patients (3.6%) after a median follow-up of 27 months (IQR, 18-38 months). The median time to LR was 13 months (IQR, 9-19 months). Sixteen of the 22 patients with LR (72.7%) were male, 14 (63.6%) received neoadjuvant chemoradiation, and 12 (54.5%) had American Joint Committee on Cancer stage III disease. Of those with LR, 16 (72.7%) had a negative circumferential radial margin and 20 (90.9%) had a negative distal resection margin, 2 (9.1%) experienced conversion to open surgery, and 15 (68.2%) also developed SR. The probability of LR-free survival at 36 months was 96% (95% CI, 94%-98%). According to the Cox proportional hazards regression model, the hazard ratio of LR was estimated to be 4.2 (95% CI, 2.9-6.2) times higher among patients with a positive circumferential radial margin than among those with a negative circumferential radial margin.Conclusions and relevanceIn this cohort study, transanal TME performed by experienced surgeons was associated with an incidence of LR and SR that is in line with the published literature on open and laparoscopic TME, suggesting that transanal TME may be an acceptable approach for management of rectal cancer.

Project description:BackgroundMinimally invasive total mesorectal excision (MiTME) and transanal total mesorectal excision (TaTME) are popular trends in mid and low rectal cancer. However, there is currently no systematic comparison between MiTME and TaTME of mid and low-rectal cancer. Therefore, we systematically study the perioperative and pathological outcomes of MiTME and TaTME in mid and low rectal cancer.MethodsWe have searched the Embase, Cochrane Library, PubMed, Medline, and Web of Science for articles on MiTME (robotic or laparoscopic total mesorectal excision) and TaTME (transanal total mesorectal excision). We calculated pooled standard mean difference (SMD), relative risk (RR), and 95% confidence intervals (CIs). The protocol for this review has been registered on PROSPERO (CRD42022374141).ResultsThere are 11010 patients including 39 articles. Compared with TaTME, patients who underwent MiTME had no statistical difference in operation time (SMD -0.14; CI -0.31 to 0.33; I2=84.7%, P=0.116), estimated blood loss (SMD 0.05; CI -0.05 to 0.14; I2=48%, P=0.338), postoperative hospital stay (RR 0.08; CI -0.07 to 0.22; I2=0%, P=0.308), over complications (RR 0.98; CI 0.88 to 1.08; I2=25.4%, P=0.644), intraoperative complications (RR 0.94; CI 0.69 to 1.29; I2=31.1%, P=0.712), postoperative complications (RR 0.98; CI 0.87 to 1.11; I2=16.1%, P=0.789), anastomotic stenosis (RR 0.85; CI 0.73 to 0.98; I2=7.4%, P=0.564), wound infection (RR 1.08; CI 0.65 to 1.81; I2=1.9%, P=0.755), circumferential resection margin (RR 1.10; CI 0.91 to 1.34; I2=0%, P=0.322), distal resection margin (RR 1.49; CI 0.73 to 3.05; I2=0%, P=0.272), major low anterior resection syndrome (RR 0.93; CI 0.79 to 1.10; I2=0%, P=0.386), lymph node yield (SMD 0.06; CI -0.04 to 0.17; I2=39.6%, P=0.249), 2-year DFS rate (RR 0.99; CI 0.88 to 1.11; I2=0%, P = 0.816), 2-year OS rate (RR 1.00; CI 0.90 to 1.11; I2=0%, P = 0.969), distant metastasis rate (RR 0.47; CI 0.17 to 1.29; I2=0%, P = 0.143), and local recurrence rate (RR 1.49; CI 0.75 to 2.97; I2=0%, P = 0.250). However, patients who underwent MiTME had fewer anastomotic leak rates (SMD -0.38; CI -0.59 to -0.17; I2=19.0%, P<0.0001).ConclusionThis study comprehensively and systematically evaluated the safety and efficacy of MiTME and TaTME in the treatment of mid to low-rectal cancer through meta-analysis. There is no difference between the two except for patients with MiTME who have a lower anastomotic leakage rate, which provides some evidence-based reference for clinical practice. Of course, in the future, more scientific and rigorous conclusions need to be drawn from multi-center RCT research.Systematic review registrationhttps://www.crd.york.ac.uk/PROSPERO, identifier CRD42022374141.