Multiple foci of dilated vessels as a new predictor of metachronous esophageal cancer.

OBJECTIVES: Prediction of the risk of esophageal squamous cell carcinoma (SCC) by endoscopic findings without iodine staining, which is irritating to the esophagus, would be beneficial. In a previous retrospective study, we found that multiple foci of dilated vascular areas (MDV) of the esophageal mucosa, seen in narrow-band imaging (NBI)/blue laser imaging (BLI), are associated with iodine-unstained lesions and, thus, may be a predictor of esophageal SCC. This prospective study aimed to investigate the association between MDV and metachronous esophageal SCC. METHODS: Patients with a history of endoscopic resection for esophageal SCC were included in the study. First, evaluation of the MDV using NBI or BLI was conducted during the initial endoscopy. The patients were then monitored for metachronous esophageal SCC by endoscopic surveillance. The association between the number of MDV and incidence of metachronous esophageal SCC was investigated. RESULTS: From February 2018 to May 2019, 206 patients were enrolled and 201 patients were included in the analysis. Patients were followed up until October 2022. The median (interquartile range) endoscopic follow-up period was 1260 (1105-1348) days. The incidence of metachronous esophageal SCC at 2 years was 7.1% in patients with MDV ≤4 and 13.9% in patients with MDV ≥5 (P < 0.01). In the multivariate analysis, MDV was an independent predictor of metachronous esophageal SCC, with an odds ratio (95% confidence interval) of 2.37 (1.06-5.31). CONCLUSION: Multiple foci of dilated vascular area is a useful predictor for stratifying the risk of metachronous esophageal SCC.

Feasibility of underwater endoscopic mucosal resection for endoscopic management of gastric neoplasms in patients with familial adenomatous polyposis.

BACKGROUND: Underwater endoscopic mucosal resection (UEMR) has been developed as an effective endoscopic intervention for colon, rectum, and duodenum neoplasms. However, there are no comprehensive reports regarding the stomach, and its safety and efficacy are unknown. We aimed to examine the feasibility of UEMR for gastric neoplasms in patients with familial adenomatous polyposis (FAP). METHODS: We retrospectively extracted data of patients with FAP who underwent endoscopic resection (ER) for gastric neoplasms at Osaka International Cancer Institute from February 2009 to December 2018. Elevated gastric neoplasms of ≤ 20 mm in diameter were extracted, and conventional endoscopic mucosal resection (CEMR) and UEMR were compared. Furthermore, outcomes after ER until March 2020 were examined. RESULTS: 91 endoscopically resected gastric neoplasms were extracted from 31 patients with 26 pedigrees, and 12 neoplasms underwent CEMR and 25 neoplasms underwent UEMR was compared. The procedure time was shorter for UEMR than for CEMR. There was no significant difference between en bloc resection and R0 resection rates by EMR methods. CEMR and UEMR showed postoperative hemorrhage rates of 8% and 0%, respectively. Residual/local recurrent neoplasms were identified in four lesions (4%), but additional endoscopic intervention (three UEMR and one cauterization) resulted in a local cure. CONCLUSION: UEMR was feasible in gastric neoplasms of FAP patients, especially in elevated lesions and those of ≤ 20 mm in diameter.

Distinct approaches (antecolic and retrocolic) according to tumor location in laparoscopic and endoscopic cooperative surgery for non-ampullary duodenum tumors.

AIM: Laparoscopic and endoscopic cooperative surgery for early non-ampullary duodenum tumors (D-LECS) is now noted because of its safety and lower invasiveness. Here, we introduce two distinct approaches (antecolic and retrocolic) according to the tumor location during D-LECS. METHODS: From October 2018 to March 2022, 24 patients (25 lesions) underwent D-LECS. Two (8%), two (8%), 16 (64%), and five (20%) lesions were located in the first portion, in the second portion to Vater's papilla, around the inferior duodenum flexure, and in the third portion of the duodenum, respectively. The median preoperative tumor diameter was 22.5 mm. RESULTS: Antecolic and retrocolic approaches were employed in 16 (67%) and 8 (33%) cases, respectively. LECS procedures, such as two-layer suturing after full-thickness dissection and laparoscopic reinforcement by seromuscular suturing after endoscopic submucosal dissection (ESD), were performed in five and 19 cases, respectively. Median operative time and blood loss were 303 min and 5 g, respectively. Intraoperative duodenal perforations occurred in three of 19 cases during ESD; however, they were successfully laparoscopically repaired. Median times until start diet and postoperative hospital stay were 4.5 and 8 days, respectively. Histological examination of the tumors revealed nine adenomas, 12 adenocarcinomas, and four GISTs. Curative resection (R0) was achieved in 21 cases (87.5%). In a comparison of the surgical short outcomes between antecolic and retrocolic approaches, there was no significant difference. CONCLUSION: D-LECS can be a safe and minimally invasive treatment option for non-ampullary early duodenal tumors, and two distinct approaches according to the tumor location are feasible.

Efficacy of a new traction method using ring-shaped thread for endoscopic submucosal dissection in the pharynx.

BACKGROUND: Endoscopic submucosal dissection (ESD) is a minimally invasive treatment for pharyngeal cancers. However, pharyngeal ESD is sometimes technically challenging because of the narrow and complex space in which to work. Traction is important to complete the procedure efficiently. Here, we report the technical details and efficacy of a new traction method for pharyngeal ESD using ring-shaped thread and grasping forceps. METHODS: We analyzed pharyngeal ESD performed between January 2016 and March 2021 at our Institute. We designated cases resected using ring-shaped threads "Group R" and those resected without ring-shaped threads as conventional "Group C", and compared the technical outcomes between them. Multivariate analysis and the inverse probability treatment weighting (IPTW) method using propensity scores were adjusted by confounding variables. RESULTS: We analyzed 89 lesions from 68 patients, of which 46 were in Group R and 43 in Group C. Median procedure time and median dissection speed were significantly shorter in Group R than C (37 min vs. 55 min, and 16.0 mm2/min vs. 7.0 mm2/min, respectively, both P < 0.05). These results were confirmed by both multivariate analysis and after IPTW adjustment. All lesions were resected en bloc, and the complete resection rate was not significantly different between Group R and C (91.3% vs. 79.1%, P = 0.14). There were no treatment-related adverse events in either group. CONCLUSIONS: The traction method using ring-shaped thread increases the efficiency of pharyngeal ESD. This simple new traction method should be a useful option for pharyngeal ESD.

Comparison of endoscopic submucosal resection with ligation and endoscopic submucosal dissection for small rectal neuroendocrine tumors: A multicenter retrospective study.

Objectives: Endoscopic submucosal resection with band ligation (ESMR-L) and endoscopic submucosal dissection (ESD) are both standard endoscopic resection methods for rectal neuroendocrine tumors (NETs) <10 mm in size. However, there is no definitive consensus on which is better. Here, we compared the efficacy of ESMR-L and ESD for small rectal NETs. Methods: This was a multicenter retrospective cohort study including 205 patients with rectal NETs who underwent ESMR-L or ESD. Treatment outcomes were compared by univariate analysis, multivariate analysis, and inverse probability treatment weighting (IPTW) using propensity scores. Subgroup analysis evaluated the impact of the endoscopist's experience on the technical outcome. Results: Eighty-nine patients were treated by ESMR-L and 116 by ESD. The R0 resection rate was not significantly different between the two (90% vs. 92%, p = 0.73). The procedure time of ESMR-L was significantly shorter than for ESD (17 min vs. 52 min, p < 0.01) and the hospitalization period was also significantly shorter (3 days vs. 5 days, p < 0.01). These results were confirmed by multivariate analysis and also after IPTW adjustment. The procedure time of ESD was significantly prolonged by a less-experienced endoscopist (49 min vs. 70 min, p = 0.02), but that of ESMR-L was not affected (17 min vs. 17 min, p = 0.27). Conclusions: For small rectal NETs, both ESMR-L and ESD showed similar high complete resection rates. However, considering the shorter procedure time and shorter hospitalization period, ESMR-L is the more efficient treatment method, especially for less-experienced endoscopists.

Mixed Neuroendocrine Non-Neuroendocrine Neoplasm Arising in the Ectopic Gastric Mucosa of Esophagus.

Esophageal neuroendocrine neoplasms are extremely rare, and their prognosis is poor. Mixed neuroendocrine non-neuroendocrine neoplasms (MiNENs) are even more rare and are defined as tumors consisting of neuroendocrine carcinoma and either adenocarcinoma or squamous cell carcinoma. We report a rare case featuring endoscopic submucosal dissection (ESD) for an esophageal MiNEN, arising from the ectopic gastric mucosa in the lower thoracic esophagus. A 92-year-old male patient was referred to this hospital for investigation of an esophageal tumor. An endoscopic examination revealed a 10 mm elevated lesion, with 8 mm flat areas on the anal side, within the ectopic gastric mucosa located in the lower thoracic esophagus. ESD was carried out, and a histopathological examination revealed a tubular adenocarcinoma composed of differentiated neuroendocrine cells. Immunohistochemical staining was positive for synaptophysin and negative for chromogranin A. The labeling index of Ki-67 was more than 80%. Based on these results, we diagnosed the lesion as an esophageal MiNEN, arising in the ectopic gastric mucosa of the esophagus. The patient remains alive, without recurrence of cancer, 24 months after ESD.

Performance of perioperative antibiotics against post-endoscopic submucosal dissection coagulation syndrome: a multicenter randomized controlled trial.

BACKGROUND AND AIMS: This study aimed to evaluate the efficacy of perioperative antibiotics against post-endoscopic submucosal dissection (ESD) coagulation syndrome (PECS) in patients undergoing colorectal ESD. METHODS: A prospective, multicenter, randomized controlled, parallel, superiority trial was conducted at 21 Japanese tertiary institutions. Patients with superficial colorectal lesions ≥20 mm and those undergoing ESD management for a single lesion were eligible. Patients with perforation during and after ESD were withdrawn. Before the ESD procedure, participants were randomly assigned (1:1) to either undergo conventional treatment (nonantibiotic group) or investigational treatment (antibiotic group). In the antibiotic group, 3 g of ampicillin-sulbactam was administered just before, 8 hours after, and the morning after ESD. The primary endpoint was the incidence of PECS. The onset of PECS was defined as localized abdominal pain (both spontaneous pain and tenderness) and fever (≥37.6°C) or inflammatory response (leukocytosis [≥10,000 cells/µL] or elevated C-reactive protein level [≥.5 mg/dL]). RESULTS: From February 5, 2019 to September 7, 2020, 432 patients were enrolled and assigned to the antibiotic group (n = 216) or nonantibiotic group (n = 216). After withdrawal of 52 patients, 192 in the antibiotic group and 188 in the nonantibiotic group were analyzed. PECS occurred in 9 of 192 patients (4.7%) in the antibiotic group and 14 of 188 patients (7.5%) in the nonantibiotic group, with an odds ratio of .61 (95% confidence interval, .23-1.56; P = .29). CONCLUSIONS: Perioperative use of antibiotics was not effective in reducing the incidence of PECS in patients undergoing colorectal ESD. (Clinical trial registration number: UMIN000035178.).

Clinical impact of gastrointestinal endoscopy on the early detection of pharyngeal squamous cell carcinoma: A retrospective cohort study.

BACKGROUND: In recent years, with the growing availability of image-enhanced gastrointestinal endoscopy, gastroenterologists have contributed to the early detection of pharyngeal squamous cell carcinomas (SCC). AIM: To clarify the clinical characteristics of pharyngeal SCCs detected by gastrointestinal endoscopy. METHODS: This is a retrospective cohort study conducted in a single-center, a university hospital in Japan. We retrospectively assessed the clinical records of 522 consecutive patients with oropharyngeal or hypopharyngeal SCC who were examined in our hospital between 2011 and 2018. The lesions were classified into two groups: Group GE (detected by gastrointestinal endoscopy) and Group non-GE (detected by means other than gastrointestinal endoscopy). The clinical characteristics were compared between the two groups. Continuous data were compared using the Mann-Whitney U test. Pearson's χ 2 test or Fisher's exact test was used to analyze the categorical data and compare proportions. The Kaplan-Meier method was used to estimate the cumulative patient survival rates. RESULTS: In our study group, the median age was 65 years and 474 patients (90.8%) were male. One hundred and ninety-six cases (37.5%) involved the oropharynx and 326 cases (62.5%) involved the hypopharynx. Three hundred and ninety-five cases (75.7%) had some symptoms at the time of diagnosis. One hundred and forty-five (27.8%) cases had concurrent ESCC or a history of ESCC. One hundred and sixty-four (31.4%) cases were detected by gastrointestinal endoscopy and classified as Group GE. The proportions of asymptomatic cases, cTis-1 cases and cases with no lymph node metastasis were significantly higher in Group GE than Group non-GE (61.6% vs 7.3%, P < 0.001, 32.9% vs 12.0%, P < 0.001 and 69.5% vs 19.0%, P < 0.001). Endoscopic laryngo-pharyngeal surgery or endoscopic submucosal dissection were performed in only 0.6% of the lesions in Group non-GE but in 21.3% of the lesions in Group GE (P < 0.001). Overall survival was significantly longer in Group GE than in Group non-GE (P = 0.018). The 2-year and 4-year survival rates were 82.5% and 70.7% in Group GE, and 71.5% and 59.0% in Group non-GE, respectively. CONCLUSION: Gastrointestinal endoscopy plays an important role in the early detection and improving the prognosis of pharyngeal SCCs.

Gastric neoplasms in patients with familial adenomatous polyposis: endoscopic and clinicopathologic features.

BACKGROUND AND AIMS: Gastric neoplasms in patients with familial adenomatous polyposis (FAP) occur at a high rate and can cause death. The endoscopic findings of gastric neoplasms in these patients are characteristic but not well recognized. To identify the relevant characteristics to enable early detection, we retrospectively investigated endoscopic findings of gastric neoplasms in patients with FAP and then compared the clinical, histopathologic, and genetic features among subgroups. METHODS: Of 234 patients with 171 pedigrees at 2 institutes, 56 cases (24%, 133 gastric neoplasms) with 44 pedigrees were examined. Immunostaining was performed for histopathologic evaluation by 1 blinded pathologist. According to the endoscopic findings, gastric neoplasms were divided into 4 types based on location (L: antrum and pylorus, UM: the rest of the stomach) and color (W: white, T: translucent, R: reddish) and their clinicopathologic features examined. RESULTS: Of the cases, 93% could be classified into a single type. Among histologic phenotypes, high-grade dysplasia was present in 26% (type L), 41% (type UM-W), 0% (type UM-T), and 22% (type UM-R). The immunologic phenotype comprised the gastric type in 69% (93% in Type UM) and the intestinal phenotype, including the mixed type, in 31% (61% in type L). Moreover, 96% of patients had concurrent duodenal neoplasms. Adenomatous polyposis coli gene status was identified in 93% of patients; the pathogenic variant was detected in 98% but did not influence any endoscopic features. CONCLUSIONS: Gastric neoplasms in patients with FAP were stratified into 4 types according to their endoscopic findings. The endoscopic phenotype was related to the histopathologic phenotype but not to germline variants.

Magnifying endoscopy with crystal violet staining for immune checkpoint inhibitor-associated colitis.

BACKGROUND AND AIM: There exists no evidence on the relationship between endoscopic and histologic findings. Furthermore, even after multiple biopsy specimens were obtained, histologic examination usually fails to show the characteristic features of immune checkpoint inhibitor-associated colitis. In this study, we explored the endoscopic and histologic findings of immune checkpoint inhibitor-associated colitis. METHODS: Patients diagnosed with immune checkpoint inhibitor-associated colitis at our hospital between March 2018 and December 2018 were retrospectively assessed. The degree of mucosal inflammation was evaluated using endoscopic inflammation grade (inactive, mild, moderate, or severe disease) and further observed using magnifying endoscopy with crystal violet staining. Pit structures were classified into three types: regularly arranged pits with circular or elliptical shape (R type), irregularly arranged pits with inhomogeneous size and morphology (IR type), and pits with reduced density or pits that partially disappeared (AD type). RESULTS: Eleven patients (median age, 71 years; range, 44-83 years) were diagnosed with immune checkpoint inhibitor-associated colitis. All characteristic histologic findings, including crypt distortion, crypt abscesses, and apoptotic bodies, were observed at sites with moderate-to-severe endoscopic inflammation but not at sites with inactive-to-mild endoscopic inflammation. Characteristic histologic features were observed in 0%, 50%, and 100% of R-type, IR-type, and AD-type mucosa, respectively. CONCLUSIONS: We revealed the possible utility of endoscopic images for selecting suitable target sites for biopsy and showed that endoscopic findings could reduce the time lag associated with tissue diagnosis and sampling errors due to biopsy.

Long-term outcomes after endoscopic submucosal dissection for differentiated-type early gastric cancer that fulfilled expanded indication criteria: A prospective cohort study.

BACKGROUND AND AIM: Endoscopic resection for early gastric cancer (EGC) is widely performed. However, there is still a paucity of strong evidence regarding long-term outcomes after endoscopic submucosal dissection (ESD) for the expanded indication criteria of the Japanese guidelines (ver. 2010). METHODS: Endoscopic submucosal dissection was performed in patients with EGC that met the expanded indication criteria: (i) cT1a, differentiated-type EGC of 2 to 5 cm, ulcer negative or (ii) cT1a, differentiated-type EGC of ≤3 cm, ulcer positive. Patients whose pathological examination fulfilled the curative resection criteria were then enrolled in this cohort study: negative vertical margin, negative lymphovascular invasion, and (i) pT1a, differentiated-type, and ulcer negative; (ii) pT1a, differentiated-type, ≤3 cm, and ulcer positive; or (iii) pT1b1 (<500-µm submucosal invasion), differentiated-type, and ≤3 cm. Patients with only a positive horizontal margin as a noncurative factor were included for follow-up. RESULTS: From September 2003 to February 2012, a total of 356 patients underwent ESD, and 214 were enrolled in the survival analysis. One hundred twenty patients (56%) had >2 cm in diameter and ulcer-negative lesions, and 94 (44%) had ≤3 cm and ulcer-positive lesions. The vital status at 5 years after ESD was confirmed in all (100%) patients. No local or metastatic recurrence was detected; however, 26 metachronous gastric cancers developed, and 1 patient died of metachronous gastric cancer. The 5-year disease-specific and overall survival rates were 99.5% (95% confidence interval [CI], 97.2%-100%) and 93.9% (95% CI, 89.8%-96.4%), respectively. CONCLUSION: ESD for EGC that fulfills the expanded criteria is feasible and shows favorable long-term outcomes.

Features of Esophageal Adenocarcinoma in Magnifying Narrow-Band Imaging.

BACKGROUND: Several endoscopic classifications for esophageal adenocarcinoma have been proposed; however, gastric adenocarcinoma is much more common than esophageal or esophagogastric junctional (EGJ) adenocarcinoma in East Asian countries. We, therefore, investigated whether an endoscopic diagnostic algorithm for gastric adenocarcinoma could be used for esophageal or EGJ adenocarcinoma. METHODS: One hundred eighteen consecutive patients who underwent endoscopic resection or surgery for intramucosal esophageal or EGJ adenocarcinoma, at the Osaka International Cancer Institute between January 2006 and December 2017, were included in this retrospective study. Their lesions were classified as Siewert type 1 or 2, and the presence of endoscopic magnifying narrow-band imaging findings for diagnosing gastric adenocarcinoma was evaluated. RESULTS: We evaluated 125 adenocarcinomas in 118 patients (29 type 1 and 96 type 2). Demarcation lines (DLs) were seen in 7 (24%) type 1 and 53 (55%) type 2 lesions. Irregular mucosal patterns were present in 2 (7%) type 1 and 22 (23%) type 2 lesions. Irregular vascular patterns were present in 26 (90%) type 1 and 50 (52%) type 2 lesions. According to the magnifying endoscopy diagnostic algorithm for gastric adenocarcinoma, only 7 (24%) type 1 and 52 (54%) type 2 lesions were correctly diagnosed as cancers (p = 0.005). CONCLUSION: The magnifying endoscopy diagnostic algorithm for gastric cancer may not be useful for esophageal or EGJ adenocarcinomas because of the low visibility of DLs, especially in Siewert type 1 adenocarcinoma.

Narrow band imaging under less-air condition improves the visibility of superficial esophageal squamous cell carcinoma.

BACKGROUND: The current virtual chromoendoscopy equipment cannot completely detect superficial squamous cell carcinoma (SCC) in the esophagus, despite its development in the recent years. Thus, in this study, we aimed to elucidate the appropriate air volume during endoscopic observation to improve the visibility of esophageal SCC. METHODS: This retrospective study included a total of 101 flat type esophageal SCCs identified between April 2017 and January 2019 at the Department of Gastrointestinal Oncology, Osaka International Cancer Institute. Video images of narrow band imaging (NBI) under both less-air and standard-air conditions were recorded digitally. Videos were evaluated by five endoscopists. Relative visibility between less-air and standard-air conditions of the brownish area, brownish color change of the epithelium, and dilated intrapapillary capillary loop (IPCL) were graded as 5 (definitely better under less-air condition) to 1 (definitely worse under less-air condition), with 3 indicating average visibility (equivalent to standard-air observation). RESULTS: The mean (standard deviation) visibility score of the brownish area, brownish color change of the epithelium, and dilated IPCLs under less-air condition were 3.94 (0.58), 3.73 (0.57), and 4.13 (0.60), respectively, which were significantly better than that under standard-air condition (p < 0.0001). Esophageal SCC evaluated as ≥ 4 in the mean visibility score of the brownish area, brownish color change of the epithelium, and dilated IPCLs accounted for 50% (51/101 lesions), 34% (34/101 lesions), and 67% (68/101 lesions), respectively. CONCLUSIONS: The present results suggested that NBI with less air might improve the visibility of flat type esophageal SCC compared with NBI with standard air. Less-air NBI observation may facilitate the detection of flat type esophageal SCC. TRIAL REGISTRATION: The present study is a non-intervention trial.

Usefulness of epinephrine-added injection solution to reduce procedure time for gastric endoscopic submucosal dissection.

Background and study aims Epinephrine-added submucosal injection solution is used to facilitate hemostasis of non-variceal upper gastrointestinal bleeding and to prevent delayed bleeding of large pedunculated colorectal lesions. However, its benefit in gastric endoscopic submucosal dissection (ESD) for early gastric cancer (EGC) is unclear. The effectiveness of epinephrine-added injection solution for outcomes of gastric ESD was examined using propensity score matching analysis. Patients and methods A total of 1,599 patients with solitary EGC (83 with non-epinephrine-added solution and 1,516 with epinephrine-added solution) between 2011 and 2018 were enrolled. Propensity scores were calculated to balance the distribution of baseline characteristics: age, sex, tumor location, specimen size, presence of ulcer scar, tumor depth, histological tumor type, and operators' experience, and 1:3 matching was performed. En bloc resection rate, mean procedure time, delayed bleeding rate, and perforation rate were compared between the non-epinephrine (n = 79) and epinephrine (n = 237) groups. Results Mean procedure time was significantly shorter in the epinephrine group than in the non-epinephrine group (60 vs. 78 min, P  < 0.001). No significant difference was found in the rate of en bloc resection (both 99 %), incidence of delayed bleeding (both 6 %), or perforation (0 vs. 0.8 %) between the two groups. In multiple linear regression analysis, use of epinephrine-added solution was independently associated with short procedure time ( P  < 0.001) after adjustment for other covariates. Conclusion The results suggest that epinephrine-added injection solution is useful for reduction of gastric ESD procedure time, warranting validation in a randomized controlled trial.

Stratification of gastric cancer risk using a deep neural network.

BACKGROUND AND AIM: Stratifying gastric cancer (GC) risk and endoscopy findings in high-risk individuals may provide effective surveillance for GC. We developed a computerized image- analysis system for endoscopic images to stratify the risk of GC. METHODS: The system was trained using images taken during endoscopic examinations with non-magnified white-light imaging. Patients were classified as high-risk (patients with GC), moderate-risk (patients with current or past Helicobacter pylori infection or gastric atrophy), or low-risk (patients with no history of H. pylori infection or gastric atrophy). After selection, 20,960, 17,404, and 68,920 images were collected as training images for the high-, moderate-, and low-risk groups, respectively. RESULTS: Performance of the artificial intelligence (AI) system was evaluated by the prevalence of GC in each group using an independent validation dataset of patients who underwent endoscopic examination and H. pylori serum antibody testing. In total, 12,824 images from 454 patients were included in the analysis. The time required for diagnosing all the images was 345 seconds. The AI system diagnosed 46, 250, and 158 patients as low-, moderate-, and high risk, respectively. The prevalence of GC in the low-, moderate-, and high-risk groups was 2.2, 8.8, and 16.4%, respectively (P = 0.0017). Three experienced endoscopists also successfully stratified the risk; however, interobserver agreement was not satisfactory (kappa value of 0.27, indicating fair agreement). CONCLUSION: The current AI system detected significant differences in the prevalence of GC among the low-, moderate-, and high-risk groups, suggesting its potential for stratifying GC risk.

Application of Convolutional Neural Networks for Detection of Superficial Nonampullary Duodenal Epithelial Tumors in Esophagogastroduodenoscopic Images.

OBJECTIVES: A superficial nonampullary duodenal epithelial tumor (SNADET) is defined as a mucosal or submucosal sporadic tumor of the duodenum that does not arise from the papilla of Vater. SNADETs rarely metastasize to the lymph nodes, and most can be treated endoscopically. However, SNADETs are sometimes missed during esophagogastroduodenoscopic examination. In this study, we constructed a convolutional neural network (CNN) and evaluated its ability to detect SNADETs. METHODS: A deep CNN was pretrained and fine-tuned using a training data set of the endoscopic images of SNADETs (duodenal adenomas [N = 65] and high-grade dysplasias [HGDs] [N = 31] [total 531 images]). The CNN evaluated a separate set of images from 26 adenomas, 8 HGDs, and 681 normal tissue (total 1,080 images). The gold standard for both the training data set and test data set was a "true diagnosis" made by board-certified endoscopists and pathologists. A detected tumor was marked with a rectangular frame on the endoscopic image. If it overlapped at least a part of the "true tumor" diagnosed by board-certified endoscopists, the CNN was considered to have "detected" the SNADET. RESULTS: The trained CNN detected 94.7% (378 of 399) of SNADETs on an image basis (94% [280 of 298] of adenomas and 100% [101 of 101] of HGDs) and 100% on a tumor basis. The time needed for screening the 399 images containing SNADETs and all 1,080 images (including normal images) was 12 and 31 seconds, respectively. DISCUSSION: We used a novel algorithm to construct a CNN for detecting SNADETs in a short time.