Expert Colonoscopists Research Articles

M1510: Computer Aided Polyp Characterisation

90% of polyps removed at colonoscopy are less than 10mm in size and approximately 50% of these are neoplastic. An ability to differentiate between hyperplastic polyps and adenomas during colonoscopy would avoid unnecessary polypectomy for hyperplastic polyps and decrease procedural ‘add on' costs by reducing histopathology and outpatient follow-up requirements. It would also enable ascertainment of surveillance intervals immediately post procedure. Expert colonoscopists have high accuracy for polyp characterisation when using narrow band imaging (NBI) but non-experts are significantly less accurate. A computer algorithm designed to characterise polyps would potentially narrow performance gaps between experts and non-experts.

Reply

Reply

A Comparison of the Diagnostic Performance of Guaiac and Immunochemical Fecal Occult Blood Tests with Colonoscopy for the Detection of Colorectal Neoplasia

Purpose: While screening for colorectal neoplasia using fecal occult blood tests (FOBTs) has demonstrated a reduction in colorectal cancer morbidity and mortality, consensus has not been reached as to the best FOBT to use in large scale population screening initiatives. Our aim was to compare the diagnostic performance characteristics of the guaiac-based (G-FOBT) Hemoccult II® (Beckman Coulter) and two fecal immunochemical tests (I-FOBT); the Hemoccult ICT® (Beckman Coulter) and MagStream HemSp® (Fujirebio) relative to colonoscopy. Methods: We used a prospective observational cohort design. 1000 asymptomatic individuals were recruited from the wait lists of 20 urban gastroenterology practices in Edmonton, Canada. Both average risk (50-75 years of age) and moderate risk patients (based on family history of colon cancer or previous history of colonic polyps) were recruited from April 2008 to May 2009. At least 1 month prior to the colonoscopy by expert colonoscopists, all subjects attended an educational session and collected three fecal occult blood tests from the same stool at home on 2 consecutive days. No dietary modifications were required. Laboratory technicians and colonoscopists were blinded to results of colonoscopy and FOBT respectively. Results: 1000 patients have been recruited into the study. Results are presented for the first 751 cases. (53.4% female, mean age 56). With colonoscopy, the adenoma detection rate was 24.7%, the advanced adenoma detection rate was 5.7% and the cancer detection rate was 0.3%. The performance characteristics of the 3 occult blood tests were compared relative to the colonoscopic findings as found in Table 1.Table 1: Sensitivity and specificity of G-FOBT and I-FOBTs for colorectal neoplasiaConclusion: Compared to G-FOBT, the I-FOBT tests were more sensitive for colonic advanced adenomas and cancers with a reasonable trade-off in specificity.

Eye Tracking Search Patterns in Expert and Trainee Colonoscopists: A Novel Method of Assessing Endoscopic Competency?

Introduction: Colonoscopy training programs stress proficiency, but objective measures of competency are limited to numbers of procedures performed, trainer recommendations and examinations using endoscopy simulators. More complex competencies, such as active visual search and pathology recognition during colonoscopy are more difficult to measure, but the use of advanced computerized eye tracking technology allows this to be performed for the first time. Aim: To evaluate visual search patterns in expert and trainee colonoscopists to determine differences associated with increasing proficiency.

Computed tomography colonography compared with conventional colonoscopy for the detection of colorectal polyps

Objective To determine the diagnostic accuracy of computed tomography colonography (CTC) compared with conventional colonoscopy (CC). Methods Patients with an indication of CC were included. Fifty patients underwent CTC using multidetector CT before diagnostic colonoscopy was performed by an expert colonoscopist. Diagnostic accuracy was assessed individually both for each polyp and for each patient. Results Fifty patients were included and 40 polyps were analyzed. The by-polyp sensitivity of CTC was 15% for polyps 5 mm or less, 75% for polyps 5-10 mm and 75% for polyps 10 mm or larger. By-patient specificity was 6% for polyps 5 mm or less, 75% for polyps 5-10 mm and 80% for polyps 10 mm or larger. The specificity of CTC was 94%. CTC was preferred over CC by 90% of the patients. The mean colonoscopy examination time was 30 minutes for CC and 35 minutes for CTC (p < 0.05). Conclusions The sensitivity of CTC is moderate in detecting polyps larger than 10 mm, low in detecting 5-10 mm polyps and very low in detecting those less than 5 mm. The overall specificity of the procedure was 94%. Procedure time was lower with CC than with CTC but the latter was better tolerated by most patients.

TECHNIQUES AND PITFALLS OF ENDOSCOPIC SUBMUCOSAL DISSECTION FOR COLORECTAL TUMORS

Endoscopic submucosal dissection (ESD) for colorectal tumors is steadily being developed. Safety and standardization of ESD for colorectal tumors have not been yet established because of the technical difficulties and the unsuitable anatomical characteristics of the colon and rectum. The authors mainly use a Flex knife for mucosal incision and a Hook knife for submucosal dissection to perform ESD safely. Skillful colonoscopic control, selection of scope, distal attachment tip hood, adequate high‐frequency generator and correct approach strategy should all be considered for safe performance of ESD. However, the incidence of indicative lesions is rare because the majority of colorectal tumors are adenomatous large laterally spreading tumors, which can be cured by intentional endoscopic piecemeal resection. At present, ESD for colorectal tumors should be performed only at central facilities that have expert colonoscopists. With the development of new devices and associated techniques, technical standardization of ESD for colorectal tumors is expected in the near future.

Colonoscopy curriculum development and performance-based assessment criteria on a computer-based endoscopy simulator.

Computer-based colonoscopy simulators (CBCS) have been developed and are being introduced into the training environment. The ability of these simulators to replicate the dimensions of patient-based diagnostic colonoscopy is good. However, the benefit of simulators to either learners or their patients has not yet been established. We describe a process by which a CBCS curriculum and CBCS-based performance criteria were established for first-year gastroenterology fellows at the Mayo Clinic in Rochester, Minnesota. We used a commercially available CBCS (AccuTouch Endoscopy Simulator, Immersion Medical, Gaithersburg, MD), which consists of a specialized colonoscope that is inserted into a computer-based module with a screen showing the colonic lumen of a virtual patient. A tutorial and six cases of varying complexity are available on the CBCS. Performance variables that are measured by the simulator include the time to complete the procedure, the distance that the scope was advanced, the degree to which the mucosa was adequately visualized, the possible complications such as colonic perforation, and the level of pain experienced by the simulated patient. To begin, we established ideal performance standards by measuring the above variables for ten "expert" faculty colonscopists who completed two cases on the CBCS. Next, we measured CBCS performance standards for five partially trained colonoscopists. Finally, two non-physician gastrointestinal assistants, without prior endoscopic training, were asked to practice on the simulator to determine the time and procedure frequency required to improve their CBCS proficiency. By calculating average performance standards within each of these three groups, we were able to estimate the number of CBCS cases and minimal performance standards for new trainees. Based on the learning curves for novice colonoscopists as well as the performances of partially trained and expert colonoscopists, we speculated that if CBCS training were to be beneficial, the benefit would most likely occur at the early stages of training. The curriculum we developed consists of viewing a one-hour, multimedia tutorial, which describes the procedure and various colonoscopy techniques. This is followed by nine hours of hands-on CBCS experience, during which time the trainee will complete approximately 25 CBCS colonoscopies. Before advancing to live-patient colonoscopies, the trainee must meet certain performance standards on specific CBCS cases. These standards include the ability to view the entire colon in less than 15 minutes with minimal pain and no complications. If effective, this new colonoscopy training curriculum should result in improved competency at patient-based colonoscopy, particularly in the early stages of training. To address this question, first-year gastroenterology fellows at Mayo Clinic have been randomized into two groups: (1) a group that experiences a tutorial with hands-on CBCS curriculum and (2) a group that experiences a tutorial only. Their performances at patient-based colonoscopy as well as surveys of patient satisfaction will be measured and analyzed to determine what, if any, benefit is provided by CBCS.

Magnetic imaging of colonoscopy: An audit of looping, accuracy and ancillary maneuvers

Background: Colonoscopy is one of the most frequently performed outpatient examinations. However, the procedure is still technically challenging, largely due to unpredictable looping during insertion. The aims of this study were (1) to assess the frequency of loop formation and types of loop during colonoscopy, (2) to determine the accuracy of the endoscopist's assessment of the anatomic location of the endoscope tip and of the presence and type of endoscope loop formation, and (3) to determine the efficacy of abdominal compression and change in patient position in promoting colonoscope advancement. Methods: Consecutive patients undergoing routine, day-case colonoscopy were studied using real-time, three-dimensional magnetic endoscope imaging. All examinations were performed by expert colonoscopists, blinded to the imager view. The endoscopist estimated the position of the colonoscope tip, assessed when and what type of loop had formed, and the efficacy of ancillary maneuvers such as hand pressure or patient position change. The magnetic imager view of each procedure was recorded and retrospectively analyzed. Results: One hundred complete colonoscopies were performed. Looping occurred in 91% with N-sigmoid (79%) and deep transverse (34%) being most common. Most loops (69%) were incorrectly diagnosed by the endoscopist. Atypical loops were more common in women than men (p = 0.025). The endoscopist's assessment of tip location was correct 85% of the time overall, but 100% in the cecum. Abdominal pressure was less effective (54 of 145 times, 37%) than patient position change (95 of 144 times, 66%) in promoting endoscope tip advancement. Conclusions: Looping occurs frequently during routine colonoscopy. Although the endoscopist's assessment of tip location is fairly accurate, the majority of colonoscope loops are incorrectly appreciated. Although used frequently, ancillary maneuvers (abdominal compression and patient position change) are effective in only 52% of attempts. (Gastrointest Endosc 2000;52:1-8.)