Deep learning-based anatomical position recognition for gastroscopic examination.

BACKGROUND: The gastroscopic examination is a preferred method for the detection of upper gastrointestinal lesions. However, gastroscopic examination has high requirements for doctors, especially for the strict position and quantity of the archived images. These requirements are challenging for the education and training of junior doctors. OBJECTIVE: The purpose of this study is to use deep learning to develop automatic position recognition technology for gastroscopic examination. METHODS: A total of 17182 gastroscopic images in eight anatomical position categories are collected. Convolutional neural network model MogaNet is used to identify all the anatomical positions of the stomach for gastroscopic examination The performance of four models is evaluated by sensitivity, precision, and F1 score. RESULTS: The average sensitivity of the method proposed is 0.963, which is 0.074, 0.066 and 0.065 higher than ResNet, GoogleNet and SqueezeNet, respectively. The average precision of the method proposed is 0.964, which is 0.072, 0.067 and 0.068 higher than ResNet, GoogleNet, and SqueezeNet, respectively. And the average F1-Score of the method proposed is 0.964, which is 0.074, 0.067 and 0.067 higher than ResNet, GoogleNet, and SqueezeNet, respectively. The results of the t-test show that the method proposed is significantly different from other methods (p< 0.05). CONCLUSION: The method proposed exhibits the best performance for anatomical positions recognition. And the method proposed can help junior doctors meet the requirements of completeness of gastroscopic examination and the number and position of archived images quickly.

Inter-observer variability of experts and trainees for the diagnosis of reflux esophagitis: Comparison of linked color imaging, blue laser imaging, and white light imaging.

OBJECTIVES: Diagnosis of reflux esophagitis according to the Los Angeles classification minimal change (LA-M) has a low inter-observer agreement. We aimed to investigate whether the inter-observer agreement of reflux esophagitis was better when expert endoscopists read the endoscopic images, or when the linked color imaging (LCI) or blue laser imaging (BLI)-bright mode was used. In addition, whether the inclusion of LA-M in the definition of reflux esophagitis affected the consistency of the diagnosis was investigated. METHODS: During upper endoscopy, endoscopic images of the gastroesophageal junction were taken using white light imaging (WLI), BLI-bright, and LCI modes. Four expert endoscopists and four trainees reviewed the images to diagnose reflux esophagitis according to the modified LA classification. RESULTS: The kappa values for the inter-observer variability for the diagnosis of reflux esophagitis were poor to fair among the experts (κ = â€Š0.22, 0.17, and 0.27 for WLI, BLI-bright, and LCI, respectively) and poor among the trainees (κ = â€Š0.18, 0.08, and 0.14 for WLI, BLI-bright, and LCI). The inter-observer variabilities for the diagnosis of reflux esophagitis excluding LA-M were fair to moderate (κ = â€Š0.42, 0.35, and 0.42 for WLI, BLI-bright, and LCI) among the expert endoscopists and moderate among the trainees (κ = 0.48, 0.43, and 0.51 for WLI, BLI-bright, and LCI). CONCLUSIONS: The inter-observer agreement for the diagnosis of reflux esophagitis was very low for both the expert endoscopists and the trainees, even using BLI-bright or LCI mode. However, when reflux esophagitis LA-M was excluded from the diagnosis of esophagitis, the degree of inter-observer agreement increased.

Evaluation of long-term adherence to oesophagogastroduodenoscopy quality indicators. / Evaluación del cumplimiento a largo plazo de los indicadores de calidad en la esofagogastroduodenoscopia.

INTRODUCTION: In a previous study we demonstrated that a simple training programme improved quality indicators of Oesophagogastroduodenoscopy (OGD) achieving the recommended benchmarks. However, the long-term effect of this intervention is unknown. The aim of this study was to assess the quality of OGDs performed 3 years after of having completed a training programme. MATERIAL AND METHODS: A comparative study of 2 cohorts was designed as follows: Group A included OGDs performed in 2016 promptly after a training programme and Group B with OGDs performed from January to March 2019, this group was also divided into 2 subgroups: subgroup B1 of Endoscopists who had participated in the previous training programme and subgroup B2 of Endoscopists who had not. The intra-procedure quality indicators proposed by ASGE-ACG were used. RESULTS: A total of 1236 OGDs were analysed, 600 from Group A and 636 from Group B (439 subgroup B1 and 197 subgroup B2). The number of complete examinations was lower in Group B (566 [94.3%] vs. 551 [86.6%]; p<0.001). A significant decrease was observed in nearly all quality indicators and they did not reach the recommended benchmarks: retroflexion in the stomach (96% vs. 81%; p<0.001); Seattle biopsy protocol (86% vs. 50%; p=0.03), description of the upper GI bleeding lesion (100% vs. 62%; p<0.01), sufficient intestinal biopsy specimens (at least 4) in suspected coeliac disease (92.5% vs. 18%; p<0.001), photo documentation of the lesion (94% vs. 90%; p<0.05). Regarding the overall assessment of the procedure (including correct performance and adequate photo documentation), a significant decrease was also observed (90.5% vs. 62%; p<0.001). There were no differences between subgroups B1 and B2. CONCLUSIONS: The improvement observed in 2016 after a training programme did not prevail after 3 years. In order to keep the quality of OGDs above the recommended benchmarks, it is necessary to implement continuous training programmes.

Learning Effect of Diagnosing Depth of Invasion Using Non-Extension Sign in Early Gastric Cancer.

BACKGROUND: Determining the depth of invasion is important when considering therapeutic strategies for early gastric cancer (EGC). We determined the effects of learning the non-extension sign, that is, an index of T1b2 in EGC, on identifying its depth of invasion. METHODS: Endoscopic images of 40 EGC cases (20 showing positive non-extension sign on endoscopy as T1b2 and 20 showing negative non-extension sign on endoscopy as T1a-T1b1) were randomly displayed on PowerPoint. Participants read endoscopy findings (pretest) and attended a 60-min lecture on how to read the non-extension sign. Then, they read the same images using the non-extension sign as the marker (posttest). The primary endpoint was a change in accuracy rate for determining the depth of invasion before and after attending the lecture, for nonexperts (< 80%). RESULTS: Among 35 endoscopists, 12 were nonexperts; their test results were used for analyses. Accuracy rates for pretest and posttest among nonexperts were 75.2 and 82.5%, respectively, showing a significant increase in the accuracy rate after learning to read the non-extension sign (p = 0.003). CONCLUSION: Nonexperts' diagnostic ability to determine the depth of invasion of EGC improved by learning to read the non-extension sign. Thus, the non-extension sign is considered a simple and useful diagnostic marker.

Direct observation of procedural skills (DOPS) assessment in diagnostic gastroscopy: nationwide evidence of validity and competency development during training.

BACKGROUND: Validated competency assessment tools and the data supporting milestone development during gastroscopy training are lacking. We aimed to assess the validity of the formative direct observation of procedural skills (DOPS) assessment tool in diagnostic gastroscopy and study competency development using DOPS. METHODS: This was a prospective multicentre (N = 275) analysis of formative gastroscopy DOPS assessments. Internal structure validity was tested using exploratory factor analysis and reliability estimated using generalisability theory. Item and global DOPS scores were stratified by lifetime procedure count to define learning curves, using a threshold determined from receiver operator characteristics (ROC) analysis. Multivariable binary logistic regression analysis was performed to identify independent predictors of DOPS competence. RESULTS: In total, 10086 DOPS were submitted for 987 trainees. Exploratory factor analysis identified three distinct item groupings, representing 'pre-procedure', 'technical', and 'post-procedure non-technical' skills. From generalisability analyses, sources of variance in overall DOPS scores included trainee ability (31%), assessor stringency (8%), assessor subjectivity (18%), and trainee case-to-case variation (43%). The combination of three assessments from three assessors was sufficient to achieve the reliability threshold of 0.70. On ROC analysis, a mean score of 3.9 provided optimal sensitivity and specificity for determining competency. This threshold was attained in the order of 'pre-procedure' (100-124 procedures), 'technical' (150-174 procedures), 'post-procedure non-technical' skills (200-224 procedures), and global competency (225-249 procedures). Higher lifetime procedure count, DOPS count, surgical trainees and assessors, higher trainee seniority, and lower case difficulty were significant multivariable predictors of DOPS competence. CONCLUSION: This study establishes milestones for competency acquisition during gastroscopy training and provides validity and reliability evidence to support gastroscopy DOPS as a competency assessment tool.

[Multi-center empirical study of virtual gastroscopy simulator in the examination of clinical practice skills of specialist physicians].

Objective: To study the role of virtual gastroscopy simulator in the examination of clinical practice skills of gastrointestinal specialists. Methods: A multi-center empirical study was carried out. In June 2018, ninety participants from three tertiary care teaching hospitals in China, including Zhongshan Hospital of Fudan University, Shanghai Tongji Hospital and Xiangya Hospital of Central South University, took part in the examination. Participants were selected as follows: 30 specialists without any experience of gastroscopy exams, 30 with basic knowledge (gastroscopy exams <500 cases) and 30 with good skill (gastroscopy exams ≥500 cases). These 90 students participated in this empirical study after theoretical study and simulator training. Among them, 50 (55.6%) were undergraduates, 25 (27.8%) masters, and 15 (16.7%) doctors; 42 (46.7%) were residents, 39 (43.3%) attending physicians, and 9 (10.0%) deputy chief physicians. The assessment was divided into two parts. The first part was to use the simulator (GI mentor II) for ballooning within 10 minutes, with a total score of 40 points. The second part was a complete gastroscopy exam on the simulator, including preparation before gastroscopy exam, gastroscopy exam, post - exam management and related knowledge quiz. The difficulty and discrimination of the assessment plan were evaluated by the extreme grouping method parameters (difficulty P definition: <0.3 was difficult, 0.3 to 0.7 was medium, ≥0.7 was easy; discrimination degree D definition: <0 was poor, 0 to 0.2 was medium, ≥0.2 was good), and by comparing the virtual gastroscopy assessment scores of different experience, academic, and professional title groups. The score analysis of the student group was expressed using the mastering rate (the average score divided by the total score). The data of the normal distribution was represented by (mean±SD) (range), and the data of the nonnormal distribution was represented by M (P25, P75). Kruskal-Wallis non-parametric test was used for comparison between groups. Results: The total score of 90 students was (82.9±10.9) (55.8 to 99.0), and the mastering rate was 82.9%. The overall difficulty of the first and the second parts were 0.78 and 0.76, respectively, and the difficulty level was easy. The overall discrimination of the first and the second parts were 0.70 and 0.67, respectively, and the discrimination was good. The stratified analysis was performed according to the professional title. The median score was 83.2 (75.5, 89.0) in the residents, 82.5 (71.7, 93.6) in the attending physicians, and 93.5 (88.5, 99.0) in the deputy chief physicians, and the difference was statistically significant (H=6.213, P=0.045). According to the stratification analysis of academic qualifications, the median score was 81.7 (73.0, 87.5) in participants with undergraduate degree, 91.0 (79.0, 95.0) in those with master degree and 88.0 (81.7, 93.5) in those with doctor degree, whose difference was statistically significant (H=9.233, P=0.010). The stratified analysis of the second part showed that the median scores of the "operational process" part of the low, medium and high basis group were 33.0 (30.5, 36.5), 34.0 (32.0, 36.0) and 37.0 (35.0, 37.5), respectively, whose difference was statistically significant (H=7.603, P=0.022). The median scores of the "operational process" section of the resident, attending physician, and deputy chief physician were 33.0 (30.5, 35.0) points, 36.0 (34.0, 37.5) points and 37.0 (37.0, 37.5) points, respectively, whose difference was statistically significant (H=18.563, P=0.001). Conclusions: The virtual gastroscopy simulator can reflect the true level of gastroscopy exam skills of the students. The examination questions are moderately difficult, and there is a good discrimination of the exam.

A Newly Designed 3-Dimensional Printer-Based Gastric Hemostasis Simulator with Two Modules for Endoscopic Trainees (with Video).

Background/Aims: We used 3-dimensional (3D) printing technology to create a new hemostasis simulator for the stomach and investigated its efficacy and realism in endoscopic hemostasis training. Methods: A new stomach hemostasis simulator, with two hemostasis modules for hemoclipping and injection, was constructed using a 3D printer. Twenty-one endoscopists, including 11 first-year fellows (beginner group) and 10 faculty members (expert group), tested the performance of the simulator. We recorded and reviewed five training sessions and evaluated the simulator with questionnaires using a 7-point Likert scale. Results: The mean evaluation score of the expert group was 6.3±0.5 for the hemoclipping module and 6.0±0.6 for the injection module. The expert group strongly agreed that endoscopic handling in the simulator was realistic and reasonable for hemostasis training. The mean procedure time for hemoclipping was 72.7±7.1 seconds for the beginner group and 19.7±1.2 seconds for the expert group. The mean procedure time for injection was 92.1±9.8 seconds for the beginner group and 36.3±2 seconds for the expert group. The procedure time of beginner group became shorter with repetition and was significantly lower by the fifth trial. Conclusions: A new 3D-printed hemostasis simulator is capable of hemostasis training and can very effectively train beginners before they perform the procedure in patients with gastrointestinal bleeding.

Analysis of factors related to poor outcome after e-learning training in endoscopic diagnosis of early gastric cancer using magnifying narrow-band imaging.

BACKGROUND AND AIMS: An e-learning system teaching endoscopic diagnostic process for early gastric cancer using magnifying endoscopy with narrow-band imaging (M-NBI) was established, and its efficacy in improving the diagnostic performance for early gastric cancer was proven in a multicenter randomized controlled trial. The aim of this study was to clarify the difference in learning effect in each lesion characteristic. METHODS: Three hundred sixty-five participants diagnosed 40 gastric lesions based on M-NBI findings using the vessel-plus-surface classification system. The diagnosis data collected from each participant were assessed in this study. The accuracy of NBI cancer diagnosis was assessed using area under the receiver operating characteristics curve (AUC/ROC) analysis. AUC/ROCs were separately calculated in each lesion characteristic (shape and size), and the data were compared between tests 1 and 3. RESULTS: Continuous net reclassification improvement (cNRI) analysis of all lesions revealed significant improvement in reclassification when participants underwent e-learning (cNRI, 1.17; P < .01). The integrated discrimination improvement analysis demonstrated that the e-learning system improved diagnostic ability (.19; P < .01). According to the analysis depending on the lesion's characteristics, high AUC/ROCs were demonstrated in depressed and small lesions (<10 mm; .90 and .93, respectively). The cNRI analysis showed remarkable e-learning improvement in both depressed (cNRI, 1.33; P < .01) and small lesions (cNRI, 1.46; P < .01). However, no significant e-learning improvement was observed in elevated or flat lesions. CONCLUSIONS: In M-NBI education for endoscopists, a good learning outcome was obtained in depressed and small lesions, but a poor learning outcome was demonstrated in elevated and flat lesions. (Clinical trial registration number: UMIN000008569.).

Validity Evidence for Direct Observation of Procedural Skills in Paediatric Gastroscopy.

OBJECTIVES: Direct observation of procedural skills (DOPS) are competence-assessment tools in endoscopy. Formative paediatric gastroscopy DOPS were implemented into the UK curriculum in 2016 but lack validity evidence; we aimed to assess validity evidence using a recognised contemporary validity framework. METHODS: We performed a prospective UK-wide analysis of formative paediatric gastroscopy DOPS submitted to the e-Portfolio over 1 year. Internal structure validity was assessed using interitem correlations between DOPS items, average domain, and skillset scores and with the overall competency rating. Overall competence scores and mean DOPS scores were compared by trainee seniority and procedure count (discriminative validity). Receiver operating characteristic curve analysis was performed to explore if DOPS scores could be used to delineate procedural competency (consequential validity). RESULTS: A total of 157 DOPS assessments were completed by 20 trainers for 17 trainees. Strengths of correlations varied between DOPS components, with overall competency correlating most with technical-predominant items, domains and skillsets. Both the overall assessor's rating and mean DOPS scores increased with trainee seniority (P < 0.001) and lifetime procedure count (P < 0.001). Overall competency could be delineated using mean DOPS scores (area under receiver operating characteristic curve 0.95, P < 0.001), with a threshold of 3.9 providing optimal sensitivity (94.4%) and specificity (89.7%). CONCLUSIONS: Competencies in paediatric gastroscopy, as assessed using DOPS, vary in their correlation with overall competence and increase with trainee experience. Formative DOPS thresholds could be used to indicate readiness for summative assessment. Our study therefore provides evidence of internal structure, discriminative, and consequential validity in support of formative paediatric gastroscopy DOPS.

Changes in scoring of Direct Observation of Procedural Skills (DOPS) forms and the impact on competence assessment.

BACKGROUND: Direct Observation of Procedural Skills (DOPS) is an established competence assessment tool in endoscopy. In July 2016, the DOPS scoring format changed from a performance-based scale to a supervision-based scale. We aimed to evaluate the impact of changes to the DOPS scale format on the distribution of scores in novice trainees and on competence assessment. METHODS: We performed a prospective, multicenter (n = 276), observational study of formative DOPS assessments in endoscopy trainees with ≤ 100 lifetime procedures. DOPS were submitted in the 6-months before July 2016 (old scale) and after (new scale) for gastroscopy (n = 2998), sigmoidoscopy (n = 1310), colonoscopy (n = 3280), and polypectomy (n = 631). Scores for old and new DOPS were aligned to a 4-point scale and compared. RESULTS: 8219 DOPS (43 % new and 57 % old) submitted for 1300 trainees were analyzed. Compared with old DOPS, the use of the new DOPS was associated with greater utilization of the lowest score (2.4 % vs. 0.9 %; P < 0.001), broader range of scores, and a reduction in competent scores (60.8 % vs. 86.9 %; P < 0.001). The reduction in competent scores was evident on subgroup analysis across all procedure types (P < 0.001) and for each quartile of endoscopy experience. The new DOPS was superior in characterizing the endoscopy learning curve by demonstrating progression of competent scores across quartiles of procedural experience. CONCLUSIONS: Endoscopy assessors applied a greater range of scores using the new DOPS scale based on degree of supervision in two cohorts of trainees matched for experience. Our study provides construct validity evidence in support of the new scale format.

Live surgery courses: retrospective safety analysis after 11 editions.

INTRODUCTION: Live surgery is a topic of interest at every meeting or course in the field of surgery. The potential of laparoscopy for live broadcasting makes it an excellent tool in continuing medical education. Surgeons who participate in live surgeries are usually experienced, but several conditions may influence the results and safety of a procedure. OBJECTIVES: To analyze safety and outcomes in a series of patients who underwent surgery in the last 10 years in a live surgery course at our institution. SETTING: University public hospital in Spain. METHODS: Retrospective review of patients who underwent surgery during these courses from 2006 to 2016. Morbidity, mortality, and long-term results were analyzed. RESULTS: Of 107 patients, 74 (68.5%) were women, and 38 (35.2%) had revision surgery. Five had surgery during previous editions. The most performed procedures were Roux-en-Y gastric bypass (38.9%), sleeve gastrectomy (16.7%), and duodenal switch (14%). Ten cases were endoscopic procedures. Morbidity was 13% (14 cases), and 6 required early postoperative revision (5.6%). Most of the complications were Clavien types III and I. Bleeding was the most common (72.4%). There was no anastomotic leak, but 1 duodenal stump leak occurred. During follow-up, 6 patients had a surgical complication and 5 required revision surgery. There was no mortality. DISCUSSION: We found higher morbidity and reoperation rates. The conversion rate in long-term follow-up was higher. Despite the educational benefits, we should take into consideration the higher risk to our patients for future editions.

Changes in operative experience after the introduction of residency: an objective comparison of surgical and endoscopic volumes.

INTRODUCTION: The SingHealth Residency General Surgery Programme replaced the basic and advanced specialist training (AST) system in Singapore in 2010. The relative merits of both systems continue to be debated, especially the operational readiness of graduating residents. We set out to compare the operative and endoscopic volumes of graduates from the AST system with those from the first graduating batch of the residency programme during their registrar (or equivalent) years, as a reflection of procedural experience gained during traineeship. METHODS: Operative and endoscopic records of four graduating residents and seven AST system trainees were extracted from the cluster-wide Operating Theatre Management system and compared. Surgeries were analysed as registrar-level operations and their corresponding subspecialties. RESULTS: Registrars and senior residents performed a mean of 1,182 and 533 general surgical operations, respectively. Median percentage loss in operative volume was 50.6% (range 9.6%-75.5%). The mean number of total gastroscopies and colonoscopies performed by registrars (total gastroscopy, n = 819; total colonoscopy, n = 743) and senior residents (total gastroscopy, n = 376; total colonoscopy, n = 412) indicated a mean loss of 54.1% and 44.6%, respectively, in gastroscopic and colonoscopic experience. CONCLUSION: The residency programme aims to provide robust and complete surgical training. The operational readiness of its graduates is often scrutinised against that from the old system. Although a significant difference in surgical and endoscopic volumes was observed between the two trainee groups, this is only one marker of surgical experience and technical competence.

Effect of Information of Patients' Coping Style on Pregastroscopy Anxiety.

Examination of the relationship between patients' coping style, pregastroscopy information, and anxiety associated with gastroscopy in China was the aim of this study. A pretest, post-test, nonrandom assignment study with a two by two design was conducted. One hundred forty-five patients who underwent initial gastroscopy without sedation were classified into 2 groups on the basis of the coping style: information seekers or information avoiders using the Information Subscale of the Krantz Health Opinion Survey (KHOS-I). All participants were given standardized procedural information about gastroscopy as routine care. Half of each group was assigned to receive additional sensory information describing what sensation they would experience and how to cooperate to alleviate the discomfort. State anxiety assessed by the State Anxiety Scale of Spielberg's State-Trait Anxiety Inventory, blood pressure, and pulse were measured at enrollment and before gastroscopy. The information seekers and avoiders who received additional sensory information experienced significantly less state anxiety after the intervention. In contrast, the information seekers and avoiders who received standardized procedural information maintained their preintervention state anxiety level. Most patients reported their preference for sensory information. In conclusion, the provision of sensory information could significantly reduce patients' pregastroscopy anxiety regardless of patients' information coping style.

Interobserver agreement in detection of "white globe appearance" and the ability of educational lectures to improve the diagnosis of gastric lesions.

BACKGROUND: White globe appearance (WGA) refers to a small white lesion of globular shape underneath cancerous gastric epithelium that can be clearly visualized by magnifying endoscopy with narrowband imaging (M-NBI). WGA has been reported to be a novel endoscopic marker that is highly specific in differentiating early gastric cancer (GC) from low-grade adenoma, and has a significantly higher prevalence in early GCs than in noncancerous lesions. However, interobserver agreement in detecting WGA and whether training intervention can improve diagnostic accuracy are unknown. METHODS: Twenty sets of M-NBI images were examined by 16 endoscopists. The endoscopists attended a lecture about WGA, and examined the images again after the lecture. Interobserver agreement in detecting WGA in the second examination and increases in the proportion of correct diagnoses and the degree of confidence of diagnoses of cancerous lesions were evaluated. RESULTS: The kappa value for interobserver agreement in detecting WGA in the second examination was 0.735. The proportion of correct diagnoses was significantly higher in the second examination compared with the first examination when WGA was present (95.5% vs 55.4%; P < 0.001), but not when WGA was absent (61.6% vs 52.7%; P = 0.190). The proportion of correct diagnoses with a high degree of confidence was significantly higher in the second examination, both with WGA (91.1% vs 29.5%; P < 0.001) and without WGA (36.6% vs 20.5%; P = 0.031). CONCLUSIONS: The detection of WGA by endoscopists was highly reproducible. A brief educational lecture about WGA increased the proportion of correct diagnoses and the degree of confidence of diagnoses of GC with WGA.

The Efficacy of a Newly Designed, Easy-to-Manufacture Training Simulator for Endoscopic Biopsy of the Stomach.

BACKGROUND/AIMS: We developed a new endoscopic biopsy training simulator and determined its efficacy for improving the endoscopic biopsy skills of beginners. METHODS: This biopsy simulator, which presents seven biopsy sites, was constructed using readily available materials. We enrolled 40 participants: 14 residents, 11 first-year clinical fellows, 10 second-year clinical fellows, and five staff members. We recorded the simulation completion time for all participants, and then simulator performance was assessed via a questionnaire using the 7-point Likert scale. RESULTS: The mean times for completing the five trials were 417.7±138.8, 145.2±31.5, 112.7±21.9, and 90.5±20.0 seconds for the residents, first-year clinical fellows, second-year clinical fellows, and staff members, respectively. Endoscopists with less experience reported that they found this simulator more useful for improving their biopsy technique (6.8±0.4 in the resident group and 5.7±1.0 in the first-year clinical fellow group). The realism score of the simulator for endoscopic handling was 6.4±0.5 in the staff group. CONCLUSIONS: This new, easy-to-manufacture endoscopic biopsy simulator is useful for biopsy training for beginner endoscopists and shows good efficacy and realism.

Which part of a porcine stomach is suitable as an animal training model for gastric endoscopic submucosal dissection?

BACKGROUND AND STUDY AIMS: Training using an animal model is generally recommended before performing endoscopic submucosal dissection (ESD) in humans. This study aimed to clarify the suitability of an isolated porcine stomach as an animal training model for gastric ESD. MATERIALS AND METHODS: Study 1: six experienced endoscopists performed ESDs of six simulated lesions located in six locations in an isolated porcine stomach (1, lower third [L]/ greater curvature [Gre]; 2, L/ anterior wall [Ant]; 3, L/ posterior wall [Post]; 4, upper third [U]/ Gre; 5, U/Ant; 6, U/Post). The procedure times (s/cm(2)) were compared among the different lesion locations. A questionnaire (scored 1 - 5) completed by the endoscopists was used to assess the difference in the difficulty compared to the human stomach of the three steps of the ESD procedure (injection, incision and dissection). Study 2: the thicknesses of the mucosal layers were histologically evaluated among the six locations in three porcine and two human stomachs. RESULTS: Study 1: the procedure speed was significantly slower at L/Gre and U/Gre in the porcine stomachs (P = 0.0004). The questionnaire responses revealed a tendency toward similarity between the lower regions of the human and porcine stomachs at L/Ant, L/Post, U/Ant and U/Post during the dissection step (P = 0.056). In contrast, the injection and incision steps at L/Gre in the porcine stomach were shown to differ in difficultly level (both P = 0.0006). Study 2: the mucosal layers in the porcine stomachs were significantly thicker than those in the human stomachs in the lower areas, especially L/Gre. CONCLUSIONS: The difficulty of ESD and the histological features of the porcine stomachs were coincident with those of the human stomachs at L/Ant, L/Post, U/Ant and U/Post. These parts of the porcine stomach may be suitable as animal training models for gastric ESD.

Training in early gastric cancer diagnosis improves the detection rate of early gastric cancer: an observational study in China.

Few studies have analyzed the training of endoscopists in the diagnosis of early gastric cancer (EGC). This study assessed whether specific training of endoscopists improves the detection rate of EGC. The rates of detection of EGC by endoscopists at the Digestive Endoscopy Center of the Affiliated Nanfang Hospital of China Southern Medical University between January 2013 and May 2014 were retrospectively analyzed. Because some endoscopists received training in the diagnosis of EGC, beginning in September 2013, the study was divided into 3 time periods: January to September 2013 (period 1), September 2013 to January 2014 (period 2), and January to May 2014 (period 3). The rates of EGC detection during these 3 periods were analyzed. From January 2013 to May 2014, a total of 25,314 gastroscopy examinations were performed at our center, with 48 of these examinations (0.2%) detecting EGCs, accounting for 12.1% (48/396) of the total number of gastric cancers detected. The EGC detection rates by trained endoscopists during periods 1, 2, and 3 were 0.3%, 0.6%, and 1.5%, respectively, accounting for 22.0%, 39.0%, and 60.0%, respectively, of the gastric cancers detected during these time periods. In comparison, the EGC detection rates by untrained endoscopists during periods 1, 2, and 3 were 0.05%, 0.08%, and 0.10%, respectively, accounting for 3.1%, 6.0%, and 5.7%, respectively, of the gastric cancers detected during these times. After training, the detection rate by some trained endoscopists markedly increased from 0.2% during period 1 to 2.3% during period 3. Further, the use of magnifying endoscopy with narrow-band imaging (M-NBI) (odds ratio = 3.1, 95% confidence interval 2.4-4.1, P < 0.001) contributed to the diagnosis of EGC. In conclusion, specific training could improve the endoscopic detection rate of EGC. M-NBI contributed to the diagnosis of EGC.

The feasibility of using simulated targets in the stomachs of live pigs for full endoscopic submucosal dissection training.

BACKGROUND/AIMS: In endoscopic submucosal dissection (ESD) training, only a flat target lesion can usually be simulated in the normal mucosa. This study aimed to evaluate the feasibility of simulated targets in the stomachs of live pigs for complete training. METHODS: Six trained endoscopists with hands-on experience with ex vivo, isolated pig stomachs were enrolled in this pilot study. An endoscopic banding device was used to create a polyp that was snared, leaving an ulcerated lesion. This simulated target model was used to perform ESD in pigs. The en bloc resection rate, procedure time, complications, quality of resection, and participants' opinions on the simulated targets were compared with the conventional model. RESULTS: En bloc resections were achieved in all six simulated targets and six conventional models. The mean size of the resected specimens was 32.2 mm (range, 20 to 39 mm) in the simulated target group and 23.5 mm (range, 11 to 40 mm) in the conventional group. The target model had a high quality of resection and had a high satisfaction rate for margin identification and correct peripheral marking. CONCLUSIONS: Good identification of the lesion and ease of periphery marking in the target model may improve resection quality.

A classification of the verbal methods currently used to teach endoscopy.

BACKGROUND: As endoscopy does not lend itself well to assisting or exposure by the teacher, most of the teaching is, by necessity, done verbally. METHODS: The verbal teaching occurring during 19 colonoscopies and 14 gastroscopies was recorded by dictaphone and later transcribed. The resultant 53-page transcript was then analyzed using the Grounded Theory method. Teaching was compared between learners with less than one month versus more than one month of training and between teaching of colonoscopy versus gastroscopy. RESULTS: The process of iterative review and repeated testing yielded 6 types of verbal teaching: demonstration by the teacher, motor instructions, broad tips/tricks/pointers, verbal feedback, questioning, and non-procedural information. Inter-rater agreement was excellent (Fleiss's kappa = 0.76) between resident (DM), the non-medical educator (MP), and the medical teacher (MM). Overall, there was less non-procedural teaching (6.7% vs 23.7%, p = 0.01) and a trend towards more teaching moments per case (13.2 vs 7.9, p = 0.07) in the first month of the rotation compared to the later months. A greater proportion of the teaching for colonoscopy involved demonstration (13.7% vs. 2.7%, p = 0.040) and tips/tricks/pointers (26.6% vs. 12.4%, p = 0.012) compared to gastroscopy. CONCLUSIONS: We describe a means of categorizing verbal teaching in endoscopy that is simple and shows strong inter-rater agreement that will serve as a starting point for further studies aiming to improve how endoscopy is taught.

Learning curve for endoscopic submucosal dissection of gastric neoplasms.

BACKGROUND: Endoscopic submucosal dissection (ESD) is a widely accepted method for the treatment of early gastrointestinal neoplasms. OBJECTIVES: To investigate the learning curve of ESD performed by a single endoscopist focusing on developing the performance of dissection, shortening the procedure time, and preventing complications. PATIENTS AND METHODS: Records of 120 consecutive ESD procedures performed by a single endoscopist with an ESD knife from December 2007 to April 2013 were collected. For analysis of the learning curve, total procedures were divided into four periods, each comprising 30 sequential ESD procedures. Adjusted procedure time (min) was calculated as specimen area [π×long length (mm)×short length (mm)/4]÷procedure time. The parameters assessed were the en-bloc resection rate, complete resection rate, duration and speed of procedure time, and related complications. RESULTS: Procedure times were significantly longer with lesions located at the upper third of the stomach and with the specimen sizes exceeding 1500 mm. There were significant differences in the adjusted overall procedure time from the first to the third quarter (19.9±11.0 vs. 30.3±11.8, P=0.01) and to the fourth quarter (19.9±11.0 vs. 35.8±15.7, P<0.01), and from the second to the third quarter (21.1±8.3 vs. 30.3±11.8, P=0.04) and to the fourth quarter (21.1±8.3 vs. 35.8±15.7, P<0.01). CONCLUSION: ESD for gastric neoplasms can be performed with a steady speed after the experience of 60 ESD procedures with proper clinical outcomes. Further studies with different endoknives will be required for ESD operators as a reference.