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

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.

Interactive Training Program Improves Surgeon and Pathologist Comfort Level With Total Mesorectal Excision Grading for Rectal Cancer.

BACKGROUND: Total mesorectal excision for rectal cancer has been shown to decrease local recurrence and improve survival, and specimen grading is recommended as a best practice. However, specimen grading remains underutilized in the United States potentially because of the lack of surgeon and pathologist training in the technique. OBJECTIVE: This study aimed to determine whether an interactive webinar improves physician comfort with mesorectal grading. DESIGN: To test the effect of the program, participants completed a survey before and after participating. SETTINGS: Twelve Michigan Surgical Quality Collaborative hospitals volunteered to participate in a Total Mesorectal Excision Project. PARTICIPANTS: Total mesorectal excision grading training program participants were surgeons, surgery residents, pathologists, and pathology assistants from 12 hospitals. MAIN OUTCOME MEASURES: Comfort with grading total mesorectal excision specimens was our main outcome measure. Prewebinar surveys also measured familiarity, previous experience, and training in grade assignment, as well as interest in the training program. Postwebinar surveys measured webinar relevance and effectiveness as well as participant intention to use content in practice. RESULTS: A total of 34 participants completed the prewebinar survey and 28 participants completed the postwebinar survey. The postwebinar overall median comfort level with specimen grading of 3.64 was significantly higher than the prewebinar overall median comfort level of 2.94 (95% CI, 3.32-3.96 versus 95% CI 2.56-3.32; p = 0.007). When evaluated separately, both surgeons and pathologists reported significantly higher comfort levels with total mesorectal excision grading after the webinar. LIMITATIONS: Six participants did not complete the postwebinar survey. Surgery residents and pathology assistants were analyzed with practicing surgeons and pathologists. The pre- and postwebinar surveys were deidentified, so paired analysis was not possible. CONCLUSIONS: Our total mesorectal excision grading training program improved the comfort level of both surgeons and pathologists with specimen grading. Survey results also demonstrate that providers are interested in receiving training in rectal cancer specimen grading. See Video Abstract at http://links.lww.com/DCR/B766.PROGRAMA DE ENTRENAMIENTO INTERACTIVO MEJORA EL NIVEL DE COMODIDAD DEL CIRUJANO Y DEL PATÓLOGO CON LA CLASIFICACIÓN DE LA ESCISIÓN TOTAL DEL MESORRECTO PARA EL CÁNCER DE RECTO. ANTECEDENTES: Se ha demostrado que la escisión total del mesorrecto para el cáncer de recto disminuye la recurrencia local y mejora la supervivencia, y se recomienda la clasificación de la muestra como buena práctica de rutina. Sin embargo, sigue siendo poco utilizado en los Estados Unidos debido principalmente a la falta de formación en la técnica de cirujanos y patólogos. OBJETIVO: Determinar si un seminario interactivo en línea mejora la comodidad del médico con la clasificación mesorrectal. DISEO: Para probar el efecto del programa, los participantes completaron una encuesta antes y después de haber participado de la misma. MARCO: Doce hospitales en cooperación sobre la calidad quirúrgica de Michigan se ofrecieron como voluntarios para participar en el proyecto de Escisión Total de Mesorrecto. PARTICIPANTES: Los participantes del programa de entrenamiento en la clasificación de escisión total de mesorrecto fueron cirujanos, residentes de cirugía, patólogos y asistentes de patología de doce hospitales. PRINCIPALES RESULTADOS MEDIDOS: La comodidad con la clasificación de las muestras de escisión total de mesorrecto fue nuestro principal resultado de medición. Las encuestas previas al seminario en línea también midieron la familiaridad, la experiencia y entrenamiento previo en la clasificación, así como el interés en el programa de entrenamiento. Las encuestas posteriores midieron la relevancia y la eficacia del seminario web, así como la intención de los participantes de utilizar en la practica el contenido. RESULTADOS: Un total de 34 participantes completaron la encuesta previa, y 28 de ellos la completaron con posterioridad al seminario en línea.La mediana del nivel de comodidad general, posterior al seminario en línea, con respecto a la clasificación de la pieza de 3,64 fue significativamente mayor con respecto al valor de 2,94 previo al seminario (IC del 95%: 3,32 - 3,96 versus IC 2,56 - 3,32, respectivamente; valor de p = 0,007).Cuando fueron evaluados de manera separada, tanto los cirujanos como los patólogos reportaron niveles de comodidad significativamente más altos con la clasificación de escisión total de mesorrecto (TME) después del seminario en línea. LIMITACIONES: Seis participantes no completaron la encuesta posterior al seminario en línea. Los residentes de cirugía y los asistentes de patología fueron analizados conjuntamente con los cirujanos y patólogos en ejercicio, respectivamente. Las encuestas previas y posteriores al seminario en línea fueron anónimas, anulándose la identificación, por lo que no fue posible realizar un análisis por pares. CONCLUSIONES: Nuestro programa de entrenamiento en la clasificación de escisión total de mesorrecto mejoró el nivel de comodidad tanto de los cirujanos como de los patólogos con la clasificación de las muestras. Los resultados de la encuesta también demuestran que el personal involucrado está interesado en recibir capacitación en la clasificación de muestras de cáncer de recto. Consulte Video Resumen en http://links.lww.com/DCR/B766. (Traducción-Dr Osvaldo Gauto).

Update on Minimally Invasive Surgical Approaches for Rectal Cancer.

PURPOSE OF REVIEW: This review aims to clarify the current role of minimally invasive surgery in the treatment of rectal cancer, highlighting short- and long-term outcomes from the latest trials and studies. RECENT FINDINGS: Data from previous trials has been conflicting, with some failing to demonstrate non-inferiority of laparoscopic surgical resection of rectal cancer compared to an open approach and others demonstrating similar clinical outcomes. Robot-assisted surgery was thought to be a promising solution to the challenges faced by laparoscopic surgery, and even though the only randomized controlled trial to date comparing these two techniques did not show superiority of robot-assisted surgery over laparoscopy, more recent retrospective data suggests a statistically significant higher negative circumferential resection margin rate, decreased frequency of conversion to open, and less sexual and urinary complications. Minimally invasive surgery techniques for resection of rectal cancer, particularly robot-assisted, offer clear short-term peri-operative benefits over an open approach; however, current data has yet to display non-inferiority in terms of oncological outcomes.

Impact of robotic learning curve on histopathology in rectal cancer: A pooled analysis.

BACKGROUND: A beneficial impact of robotic proctectomy on circumferential resection margin (CRM) is expected due to the robot's articulating instruments in the pelvis. There are however concerns about a negative impact on the quality of total mesorectal excision (TME) due to the lack of tactile feedback. The aim of this study was to assess whether surgeons' learning curve impacted CRM and TME quality. METHODS: In a multicenter study, individual patient data of robotic proctectomy for resectable rectal cancer were pooled. Patients were stratified into two phases of surgeons' learning curve. Cumulative sum (CUSUM) analysis was used to determine the transition from learning phase (LP) to plateau phase (PP), which were compared. CRM was microscopically measured in mm by pathologists. TME quality was classified by pathologists as complete, nearly complete or incomplete. T-test and Chi-squared tests were used to compare continuous and categorical variables, respectively. RESULTS: 235 patients underwent robotic proctectomy by five surgeons. 83 LP patients were comparable to 152 PP patients for age (p = 0.20), gender (67.5% vs. 65.1% males; p = 0.72), BMI (p = 0.82), cancer stage (p = 0.36), neoadjuvant chemoradiation (p = 0.13), distance of tumor from anal verge (5.8 ± 4.4 vs. 5.5 ± 3.3; p = 0.56). CRM did not differ (7.7 ± 11.4 mm vs. 8.4 ± 10.3 mm; p = 0.62). The rate of complete TME quality was significantly improved in PP patients as compared to LP patients (73.5% vs. 92.1%; p < 0.001). CONCLUSION: While learning had no impact on circumferential resection margins, the quality of TME significantly improved during surgeons' plateau phase as compared to their learning phase.

Locoregional recurrences after transanal total mesorectal excision of rectal cancer during implementation.

BACKGROUND: Transanal total mesorectal excision (TaTME) has been proposed as an approach in patients with mid and low rectal cancer. The TaTME procedure has been introduced in the Netherlands in a structured training pathway, including proctoring. This study evaluated the local recurrence rate during the implementation phase of TaTME. METHODS: Oncological outcomes of the first ten TaTME procedures in each of 12 participating centres were collected as part of an external audit of procedure implementation. Data collected from a cohort of patients treated over a prolonged period in four centres were also collected to analyse learning curve effects. The primary outcome was the presence of locoregional recurrence. RESULTS: The implementation cohort of 120 patients had a median follow up of 21·9 months. Short-term outcomes included a positive circumferential resection margin rate of 5·0 per cent and anastomotic leakage rate of 17 per cent. The overall local recurrence rate in the implementation cohort was 10·0 per cent (12 of 120), with a mean(s.d.) interval to recurrence of 15·2(7·0) months. Multifocal local recurrence was present in eight of 12 patients. In the prolonged cohort (266 patients), the overall recurrence rate was 5·6 per cent (4·0 per cent after excluding the first 10 procedures at each centre). CONCLUSION: TaTME was associated with a multifocal local recurrence rate that may be related to suboptimal execution rather than the technique itself. Prolonged proctoring, optimization of the technique to avoid spillage, and quality control is recommended.

ANTECEDENTES: La escisión total del mesorrecto por vía transanal (Transanal Total Mesorectal Excision, TaTME) se ha propuesto como abordaje quirúrgico en pacientes con cáncer de recto medio e inferior. La técnica TaTME se ha introducido en los Países Bajos mediante un proceso de formación estructurado que incluye la supervisión. Este estudio evaluó el porcentaje de recidiva local durante la fase de implementación de TaTME. MÉTODOS: Se recogieron los resultados oncológicos de los primeros 10 procedimientos realizados mediante TaTME en cada uno de los 12 centros participantes como parte de una auditoría externa de implementación del procedimiento. Se reunió una cohorte más amplia de pacientes procedentes de 4 centros para analizar los efectos de la curva de aprendizaje. El criterio de valoración principal fue la presencia de recidiva locorregional. RESULTADOS: La cohorte de implementación de 120 pacientes tuvo una mediana de seguimiento de 21,9 meses. Los resultados a corto plazo incluyeron una tasa del margen de resección circunferencial positivo del 5% y una tasa de fuga anastomótica del 17,4%. La tasa global de recidiva local en la cohorte de implementación fue del 10% (12/120) con un intervalo medio de recidiva de 15,2 (DE 7) meses. El patrón de recidiva local fue multifocal en 8 de 12 casos (67%). En la cohorte ampliada (n = 266), la tasa global de recidiva fue del 5,6% (4,0%, excluyendo a los primeros 10 pacientes). CONCLUSIÓN: TaTME se asoció con un porcentaje de recidiva local multifocal que puede relacionarse con una ejecución subóptima, más que con la técnica en sí. Se recomienda una supervisión prolongada, la optimización de la técnica para evitar la diseminación tumoral, así como un control de calidad.

Structured training pathway and proctoring; multicenter results of the implementation of transanal total mesorectal excision (TaTME) in the Netherlands.

BACKGROUND: Transanal total mesorectal excision (TaTME) is a new complex technique with potential to improve the quality of surgical mesorectal excision for patients with mid and low rectal cancer. The procedure is technically challenging and has shown to be associated with a relative long learning curve which might hamper widespread adoption. Therefore, a national structured training pathway for TaTME has been set up in the Netherlands to allow safe implementation. The aim of this study was to monitor safety and efficacy of the training program with 12 centers. METHODS: Short-term outcomes of the first ten TaTME procedures were evaluated in 12 participating centers in the Netherlands within the national structured training pathway. Consecutive patients operated during and after the proctoring program for rectal carcinoma with curative intent were included. Primary outcome was the incidence of intraoperative complications, secondary outcomes included postoperative complications and pathological outcomes. RESULTS: In October 2018, 12 hospitals completed the training program and from each center the first 10 patients were included for evaluation. Intraoperative complications occurred in 4.9% of the cases. The clinicopathological outcome reported 100% for complete or nearly complete specimen, 100% negative distal resection margin, and the circumferential resection margin was positive in 5.0% of patients. Overall postoperative complication rate was 45.0%, with 19.2% Clavien-Dindo ≥ III and an anastomotic leak rate of 17.3%. CONCLUSIONS: This study shows that the nationwide structured training program for TaTME delivers safe implementation of TaTME in terms of intraoperative and pathology outcomes within the first ten consecutive cases in each center. However, postoperative morbidity is substantial even within a structured training pathway and surgeons should be aware of the learning curve of this new technique.

Defining the learning curve for transanal total mesorectal excision for rectal adenocarcinoma.

BACKGROUND: Early observational data suggest that this approach is safe and feasible, but it is technically challenging and the learning curve has not yet been determined. The objective of this study was to determine the number of cases required achieve proficiency in transanal total mesorectal excision (TA-TME) for rectal adenocarcinoma. METHODS: All TA-TME cases performed from 03/2012-01/2017 at a single high-volume tertiary care institution for rectal adenocarcinoma were included. A cumulative summation (CUSUM) analysis was performed to determine the number of cases required to reach proficiency, defined as high-quality TME (complete or near-complete mesorectal envelope, negative distal (DRM), and circumferential resection (> 1 mm; CRM) margin). The acceptable and unacceptable rates of good quality TME were defined based on the incidence of high-quality TME in laparoscopic (unacceptable rate = 81.7%) and open (acceptable rate = 86.9%) arms of the ACOSOG Z6051 trial. RESULTS: A total of 87 consecutive cases were included with mean tumor height 4.8 cm (SD 2.7) and 80% (70/87) received neoadjuvant chemoradiation. Post-operative morbidity occurred in 44% (38/87) of cases, including 21% (18/87) readmissions. Median length of stay was 4 days [IQR 3-8]. A good quality TME was performed in 95% (83/87) of cases including 98% (85/87) negative CRM, 99% (86/87) negative DRM, and 99% (86/87) complete or near-complete mesorectal envelope. CUSUM analysis reported that the good quality TME rate reaches an acceptable rate after 51 cases overall, and 45 cases if abdominoperineal resections are excluded. CONCLUSION: TA-TME is a complex technique that requires a minimum of 45-51 cases to reach an acceptable incidence of high-quality TME and lower operative duration.

Teaching robotic rectal cancer surgery at your workplace: does the presence of visiting surgeons in the operating room have a detrimental effect on outcomes?

BACKGROUND: Surgery demonstration (SD) is considered to be a mainstay of surgical education, but controversy exists concerning the patient's safety. Indeed, the presence of visiting surgeons is a source of distraction and may have an impact on surgeon's performance. This study's objective was to evaluate possible differences in outcomes between robotic sphincter-saving rectal cancer surgery (RRCS) performed during routine surgical practice versus in the presence of visiting surgeons in the operating room (OR) with direct access to the surgeon. METHODS: Retrospective case-matched studies were conducted from a prospectively collected database. 114 patients (38 with the presence of visiting surgeons) who underwent RRCS between January 2013 and September 2018 were included. Patients were matched in a 1:2 basis after propensity score analysis using five criteria: gender, body mass index, preoperative chemoradiation, type of mesorectum excision, and synchronous liver metastasis. RESULTS: There was no difference between the two groups with regard to mean operating time, estimated blood loss, conversion, and hospital stay. Also, overall (44% vs. 40%; P = 0.6), major morbidity (26% vs. 19%; P = 0.5), and unplanned reoperation (17% vs. 15%; P = 1.0) rates were not statistically different. No difference was noted with regard to the quality of mesorectum excision, or positive rate of circumferential and distal longitudinal resection margins. The mean number of harvested lymph nodes (17 vs. 14.5; P = 0.04) was lower in the SD group and the number of patients with < 12 harvested lymph nodes (31% vs. 16%; P = 0.09) was greater after SD although it did not reach statistical significance. No differences were observed in disease-free or overall survival. CONCLUSIONS: The presence of visiting surgeons in the OR seems not to interfere in the quality of rectal resection and does not compromise patient's short-term outcome and survival. However, mild differences in the extent of lymphadenectomy were observed and the surgeons performing SD may be aware of this.

The value of force and torque measurements in transanal total mesorectal excision (TaTME).

BACKGROUND: Transanal total mesorectal excision (TaTME) is associated with a relatively long learning curve. Force, motion, and time parameters are increasingly used for objective assessment of skills to enhance laparoscopic training efficacy. The aim of this study was to identify relevant metrics for accurate skill assessment in more complex transanal purse-string suturing. METHODS: A box trainer was designed for TaTME and equipped with two custom made multi-DOF force/torque sensors. These sensors measured the applied forces in the axial direction of the instruments (Fz), instrument load orientation expressed in torque (Mx and My) on the entrance port, and the full tissue interaction force (Fft) at the intestine fixation point. In a construct validity study, novices for TaTME performed a purse-string suture to investigate which parameters can be used best to identify meaningful events during tissue manipulation and instrument handling. RESULTS: Significant differences exist between pre- and post-training assessment for the mean axial force at the entrance port Fz (p = 0.01), mean torque in the entrance port Mx (p = 0.03) and mean force on the intestine during suturing Fft (p = 0.05). Furthermore, force levels during suturing exceed safety threshold values, potentially leading to dangerous complications such as rupture of the rectum. CONCLUSIONS: Forces and torque measured at the entrance port, and the tissue interaction force signatures provide detailed insight into instrument handling, instrument loading, and tissue handling during purse-string suturing in a TaTME training setup. This newly developed training setup for single-port laparoscopy that enables objective feedback has the potential to enhance surgical training in TaTME.

The effect of proctoring on the learning curve of transanal minimally invasive surgery for local excision of rectal neoplasms.

BACKGROUND: The current method of choice for local resection of benign and selected malignant rectal tumors is transanal endoscopic microsurgery. Transanal minimally invasive surgery (TAMIS) yields similar oncological results and better patient reported outcomes when compared to transanal endoscopic micro surgery. However, due to the technical complexity of TAMIS, a significant learning curve has been suggested. Data on the surgical learning curve are limited. The aim of our study was to investigate surgeon specific learning curves for TAMIS procedures for the local excision of selected rectal tumors, and analyze the effects of proctoring on operating time and outcome. METHODS: The current study was prospective of all TAMIS procedures performed by two surgeons from October 2010 to November 2017. Margin positivity, specimen fragmentation, adverse events and operative time were evaluated with a cumulative sum analysis to determine the number of procedures required to reach proficiency. Cumulative sum (CUSUM) analysis was used to determine trends in changes over time. RESULTS: The earliest adopter, surgeon A, performed 103 procedures, was not proctored and developed the standardized institutional program. Surgeon B, performed 26 cases, had the benefit of a proctorship and availability of a standardized program. The CUSUM curve for operative time showed a change after 36 cases for surgeon A and after 10 cases for surgeon B. For margin positivity proficiency was reached after 31 and 6 cases for surgeon A and B, respectively. The complications curve for surgeon A showed a three-phase learning curve with a decrease after the 26th case whereas surgeon B only had one (3.8%) complication in the learning phase with no change point in the CUSUM curve. Comparing pre- and post-proficiency periods there was a decrease in operating time for both surgeon A (84.4 ± 47.3 to 55.9 ± 30.1 min) and surgeon B (90.6 ± 64.to 53 ± 26.5 min; p < 0.001). Overall margin positivity rates decreased non significantly from 21.7 to 4.8% (p = 0.23). Complications were higher in the pre-proficiency period (21.7% vs. 13.0%; p = 0.02). Surgeon A had significantly more postoperative complications in pre-proficiency phase when compared to surgeon B (25% vs. none, p < 0.001), in the post-proficiency phase there was no statistically significant difference between both surgeons (p = 0.08). CONCLUSIONS: Our results suggest that to reach satisfactory results for TAMIS, 18-31 procedures are required. Standardized institutional operative protocols together with proficient proctorship may contribute to a shorter learning curve with fewer cases (6-10) required to reach proficiency.

Using cadaveric simulation to introduce the concept and skills required to start performing transanal total mesorectal excision.

AIM: The aim was to document the outcomes of surgeons attending a cadaveric simulation course designed to provide an introduction to transanal total mesorectal excision (TaTME). METHOD: This was a prospective observational study documenting the outcomes from classroom and wet lab activities. Follow-up questionnaires were used to monitor clinical activity after the course. RESULTS: Outcomes of 65 delegates from 12 different countries attending seven cadaveric simulation courses are described. Median time to insert and close the rectal purse-string was 15 min (range 7-50 min) and median time to complete the transanal mesorectal dissection was 105 min (range 60-260 min). Objective assessment of specimen quality showed that 42% of specimens were complete, 47% nearly complete and 11% were incomplete. Failure of the intraluminal rectal purse-string was the most common difficulty encountered. Within 6 months of attending the course, nearly half (26/55; 47%) of the surgeons who responded had performed between 1 and 13 TaTMEs. Only 8/26 (31%) of the surgeons had arranged mentoring for their first case. CONCLUSION: This training model provides high levels of trainee satisfaction and the knowledge and technical skills to enable them to start performing TaTME. There is still work to do to provide adequate supervision and mentorship for surgeons early on their learning curve that is essential for the safe introduction of this new technique.

Colorectal Surgery Fellowship Improves In-hospital Mortality After Colectomy and Proctectomy Irrespective of Hospital and Surgeon Volume.

BACKGROUND: General surgery residents are increasingly pursuing sub-specialty training in colorectal (CR) surgery. However, the majority of operations performed by CR surgeons are also performed by general surgeons. This study aimed to assess in-hospital mortality stratified by CR training status after adjusting for surgeon and hospital volume. METHODS: The Statewide Planning and Research Cooperative system database was used to identify all patients who underwent colectomy/proctectomy from January 1, 2000, to December 31, 2014, in the state of New York. Operations performed by board-certified CR surgeons were identified. The relationships between CR board certification and in-hospital mortality, in-hospital complications, length of stay, and ostomy were assessed using multivariate regression models. RESULTS: Two hundred seventy thousand six hundred eighty-four patients underwent colectomy/proctectomy over the study period. Seventy-two thousand two hundred seventy-nine (26.7%) of operations were performed by CR surgeons. Without adjusting for hospital and surgeon volume, in-hospital mortality was lower for those undergoing colectomy/proctectomy by a CR surgeon (OR 0.49, CI 0.44-0.54, p = 0.001). After controlling for hospital and surgeon volume, the odds of inpatient mortality after colectomy/proctectomy for those operated on by CR surgeons weakened to 0.76 (CI 0.68-0.86, p = 0.001). Hospital and surgeon volume accounted for 53% of the reduction in in-hospital mortality when CR surgeons performed colectomy/proctectomy. Patients who underwent surgery by a CR surgeon had a shorter inpatient stay (0.8 days, p = 0.001) and a decreased chance of colostomy (OR 0.86, CI 0.78-0.95, p < 0.001). CONCLUSIONS: For patients undergoing colectomy/proctectomy, in-hospital mortality decreased when the operation was performed by a CR surgeon even after accounting for hospital and surgeon volume.