Development and validation of artificial neural networks model for detection of Barrett's neoplasia: a multicenter pragmatic nonrandomized trial (with video).

BACKGROUND AND AIMS: The aim of this study was to develop and externally validate a computer-aided detection (CAD) system for the detection and localization of Barrett's neoplasia and assess its performance compared with that of general endoscopists in a statistically powered multicenter study by using real-time video sequences. METHODS: In phase 1, the hybrid visual geometry group 16-SegNet model was trained by the use of 75,198 images and videos (96 patients) of neoplastic and 1,014,973 images and videos (65 patients) of nonneoplastic Barrett's esophagus. In phase 2, image-based validation was performed on a separate dataset of 107 images (20 patients) of neoplastic and 364 images (14 patients) of nonneoplastic Barrett's esophagus. In phase 3 (video-based external validation) we designed a real-time video-based study with 32 videos (32 patients) of neoplastic and 43 videos (43 patients) of nonneoplastic Barrett's esophagus from 4 European centers to compare the performance of the CAD model with that of 6 nonexpert endoscopists. The primary endpoint was the sensitivity of CAD diagnosis of Barrett's neoplasia. RESULTS: In phase 2, CAD detected Barrett's neoplasia with sensitivity, specificity, and accuracy of 95.3%, 94.5%, and 94.7%, respectively. In phase 3, the CAD system detected Barrett's neoplasia with sensitivity, specificity, negative predictive value, and accuracy of 93.8%, 90.7%, 95.1%, and 92.0%, respectively, compared with the endoscopists' performance of 63.5%, 77.9%, 74.2%, and 71.8%, respectively (P < .05 in all parameters). The CAD system localized neoplastic lesions with accuracy, mean precision, and mean intersection over union of 100%, 0.62, and 0.54, respectively, when compared with at least 1 of the expert markings. The processing speed of the CAD detection and localization were 5 ms/image and 33 ms/image, respectively. CONCLUSION: To our knowledge, this is the first study describing external (multicenter) validation of AI algorithms for the detection of Barrett's neoplasia on real-time endoscopic videos. The CAD system in this study significantly outperformed nonexpert endoscopists on real-time video-based assessment, achieving >90% sensitivity for neoplasia detection. This result needs to be validated during real-time endoscopic assessment.

Predictors of long-term outcomes of endoscopic submucosal dissection of early gastric neoplasia in the West: a multicenter study.

BACKGROUND: This study aimed to determine long-term outcomes of gastric endoscopic submucosal dissection (ESD) in Western settings based on the latest Japanese indication criteria, and to examine predictors of outcomes and complications. METHODS: Data were collected from consecutive patients undergoing gastric ESD at four participating centers from 2009 to 2021. Retrospective analysis using logistic regression and survival analysis was performed. RESULTS: 415 patients were included (mean age 71.7 years; 56.4 % male). Absolute indication criteria (2018 guideline) were met in 75.3 % of patients. Median follow-up was 52 months. Post-resection histology was adenocarcinoma, high grade dysplasia, and low grade dysplasia in 49.9 %, 22.7 %, and 17.1 %, respectively. Perforation, early and delayed bleeding occurred in 2.4 %, 4.3 %, and 3.4 %, respectively. Rates of en bloc and R0 resection, and recurrence on first endoscopic follow-up were 94.7 %, 83.4 %, and 2.7 %, respectively. Relative indication (2018 guideline) for ESD was associated with R1 outcome (P = 0.02). Distal location (P = 0.002) and increased procedure time (P = 0.04) were associated with bleeding, and scarring (P = 0.009) and increased procedure duration (P = 0.003) were associated with perforation. Recurrence-free survival at 2 and 5 years was 94 % and 83 %, respectively. CONCLUSION: This is the largest Western multicenter cohort and suggests that gastric ESD is safe and effective in the Western setting. A quarter of patients fell outside the new absolute indications for ESD, suggesting that Western practice involves more advanced lesions. We identified the predictors of complications, which should help to inform future Western practice and research.

Clinical outcome of non-curative endoscopic submucosal dissection for early colorectal cancer.

OBJECTIVE: Endoscopic submucosal dissection (ESD) in a curative intent for submucosa-invasive early (T1) colorectal cancers (T1-CRCs) often leads to subsequent surgical resection in case of histologic parameters indicating higher risk of nodal involvement. In some cases, however, the expected benefit may be offset by the surgical risks, suggesting a more conservative approach. DESIGN: Retrospective analysis of consecutive patients with T1-CRC who underwent ESD at 13 centres ending inclusion in 2019 (n=3373). Cases with high risk of nodal involvement (non-curative ESD: G3, submucosal invasion>1000 µm, lymphovascular involvement, budding or incomplete resection/R1) were analysed if follow-up data (endoscopy/imaging) were available, regardless of the postendoscopic management (follow-up vs surgery) selected by the multidisciplinary teams in these institutions. Comorbidities were classified according to Charlson Comorbidity Index (CCI). Outcomes were disease recurrence, death and disease-related death rates in the two groups. Rate of residual disease (RD) at both the previous resection site and regional lymph nodes was assessed in the surgical cases as well as from follow-up in the follow-up group. RESULTS: Of 604 patients treated by colorectal ESD for submucosally invasive cancer, 207 non-curative resections (34.3%) were included (138 male; mean age 67.6±10.9 years); in 65.2% of cases, no complete resection was achieved (R1). Of the 207 cases, 60.9% (n=126; median CCI: 3; IQR: 2-4) underwent surgical treatment with RD in 19.8% (25/126), while 39.1% (n=81, median CCI: 5; IQR: 4-6) were followed up by endoscopy in all cases. Patients in the follow-up group had a higher overall mortality (HR=3.95) due to non-CRC causes (n=9, mean survival after ESD 23.7±13.7 months). During this follow-up time, tumour recurrence and disease-specific survival rates were not different between the groups (median follow-up 30 months; range: 6-105). CONCLUSION: Following ESD for a lesion at high risk of RD, follow-up only may be a reasonable choice in patients at high risk for surgery. Also, endoscopic resection quality should be improved. TRIAL REGISTRATION NUMBER: NCT03987828.

Future of Endoscopy: Brief review of current and future endoscopic resection techniques for colorectal lesions.

Endoscopic treatment of colorectal lesions has seen major developments in the last decade. It is now considered curative for intramucosal and superficial submucosal cancers. Endoscopic Mucosal Resection in expert hands has very good outcomes with low complication rates but recurrence and inadequate treatment of early cancers remain an issue. This has led to a technical evolution that can lead to one piece resection of neoplasia. This includes a range of techniques from knife assisted snare resection (KAR), endoscopic submucosal dissection (ESD) to full thickness resections. This article reviews all the resection techniques and the evidence base behind them.

Performance of a novel computer-aided diagnosis system in the characterization of colorectal polyps, and its role in meeting Preservation and Incorporation of Valuable Endoscopic Innovations standards set by the American Society of Gastrointestinal Endoscopy.

Background and aims: There has been an increasing role of artificial intelligence (AI) in the characterization of colorectal polyps. Recently, a novel AI algorithm for the characterization of polyps was developed by NEC Corporation (Japan). The aim of our study is to perform an external validation of this algorithm. Methods: The study was a video-based evaluation of the computer-aided diagnosis (CADx) system. Patients undergoing colonoscopy were recruited to record videos of colonic polyps. The frozen polyp images extracted from these videos were used for real-time histological prediction by the endoscopists and by the CADx system, and the results were compared. Results: A total of 115 polyp images were extracted from 66 patients. Sensitivity, negative predictive value and accuracy for diminutive polyps on white light imaging (WLI) and image-enhanced endoscopy (IEE) when assessed by CADx was 90.9% [95% confidence interval (CI) 77.3-100] and 95.8% [95% CI 87.5-100], 80% [95% CI 44.4-97.5] and 90.9% [95% CI 58.7-99.8], 84.8% [95% CI 72.7-97] and 84.6% [95%CI 71.8-94.9], respectively, compared to 48.1% [95%CI 37.7-59.1] and 72% [95% CI 62.5-81], 37.5% [95% CI 28.8-46.8] and 55% [95% CI 44.7-65.0], 53.7% [95% CI 44.2-63.2] and 66.7% [95% CI 59.7-73.3] when assessed by endoscopists. Concordance between histology and CADx-based post-polypectomy surveillance intervals was 93.02% on WLI and 96% on IEE. Conclusion: AI-based optical diagnosis is promising and has the potential to be better than the performance of general endoscopists. We believe that AI can help make real-time optical diagnoses of polyps meeting the Preservation and Incorporation of Valuable endoscopic Innovations standards set by the American Society of Gastrointestinal Endoscopy.

Safety and feasibility of dobutamine stress cardiac magnetic resonance for cardiovascular assessment prior to renal transplantation.

AIMS: Current guidelines recommend cardiovascular risk assessment prior to renal transplantation. There is currently no evidence for the role of cardiovascular magnetic resonance (CMR) in this population, despite an established evidence base in the non-chronic kidney disease (CKD) population. Our aim is to determine the feasibility and safety of dobutamine stress CMR (DSCMR) imaging in the risk stratification of CKD patients awaiting renal transplantation. METHODS: CKD patients who were deemed at high risk for coronary artery disease (CAD) and awaiting renal transplantation underwent DSCMR. RESULTS: Forty-one patients whose median age was 56 years (range 28­73 years) underwent DSCMR. Nineteen were undergoing haemodialysis, 10 peritoneal dialysis and 12 pre-dialysis. The aetiology of the renal failure was diabetes mellitus in 29%, glomerulonephritis in 24%, hypertension in 22% and autosomal dominant polycystic kidney disease in 10%. Thirty-eight patients (93%) achieved the end point, either positive for ischaemia or negative, achieving at least 85% of age-predicted heart rate. Two of them did not achieve target heart rate and one was discontinued because of severe headache. Of the 38 patients who achieved the end point, 35 (92%) were negative for inducible wall motion abnormalities and four (10%) were positive. There were no serious adverse effects. CONCLUSION: DSCMR is a well tolerated and viable investigation for the cardiovascular risk stratification of high-risk CKD patients prior to renal transplantation. DSCMR already has an established evidence base in the non-CKD population with superiority over other noninvasive techniques. Larger studies with outcome data are now required to define its true utility in the CKD population.