Search

Background and Aims Endoscopic differential diagnoses of gastric mucosal lesions (benign gastric ulcer, early gastric cancer [EGC], and advanced gastric cancer) remain challenging. We aimed to develop and validate convolutional neural network-based artificial intelligence (AI) models: lesion detection (AI-LD), differential diagnosis (AI-DDx), and invasion-depth (AI-ID, pT1a vs. pT1b among EGC) models. Methods This study included 1,366 consecutive patients with gastric mucosal lesions from 2 referral centers in Korea. One representative endoscopic image from each patient was used. Histological diagnoses were set as the criterion standard. The performances of the AI-DDx (training/internal/external validation set, n=1009/112/245) and AI-ID (training/internal/external validation set, n=620/68/155) were compared with visual diagnoses by independent endoscopists (stratified by novice [ 5 years of experience]) and endoscopic ultrasonography (EUS) results, respectively. Results The AI-DDx showed good diagnostic performance for both internal (area under of the receiver operating characteristic curve [AUROC]=0.86) and external validation (AUROC=0.86). The performance of the AI-DDx was better than that of the novice (AUROC=0.82, P=0.01) and intermediate endoscopists (AUROC=0.84, P=0.02), but was comparable to the experts (AUROC=0.89, P=0.12) in the external validation set. The AI-ID showed fair performances in both internal (AUROC=0.78) and external validation sets (AUROC=0.73), which were significantly better than EUS results performed by experts (internal validation: AUROC=0.62, external validation: AUROC=0.56; both P Conclusion The AI-DDx was comparable with experts and outperformed novice and intermediate endoscopists for the differential diagnosis of gastric mucosal lesion. The AI-ID performed better than EUS for the invasion-depth evaluation ( https://aiscopeseoul.com/ ).

Background and Aims Differentiated minute submucosal cancers measuring ≤3 cm that exhibit no lymphovascular invasion are considered eligible for endoscopic resection; however, the ≤3-cm criterion remains debatable. The aim of this study was to verify the optimal tumor size cutoff for endoscopic resection eligibility, and to determine whether 2-dimensional tumor size measurement would be more accurate for this purpose. Methods The clinicopathologic data for 574 patients undergoing curative surgery for differentiated minute submucosal cancer were reviewed retrospectively. A receiver operating curve analysis and bootstrapped samples were used to identify the optimal cutoff values for predicting lymph node (LN) metastasis. Results Four hundred fourteen patients were eligible for the study. LN metastasis accompanied 2.5% of differentiated minute submucosal cancers that were ≤3 cm in size and lacked lymphovascular invasion. There was no lymph node metastasis in the tumors with 1-dimensional sizes ≤1.0 cm and 2-dimensional sizes ≤1.0 cm 2 . When 10,000 bootstrapped data were used, optimal cutoff values were 2.9 cm (LN metastasis: 2.1%) and 8.3 cm 2 (LN metastasis: 2.5% incidence) for 1- and 2-dimensional tumor size, respectively. Although the area under the curve was 0.611 ( P = .046) for 2-dimensional size, specificity and accuracy were greater for the dichotomized 2-dimensional measure than for the dichotomized 1-dimensional measure ( P Conclusions One- and 2-dimensional cutoff values of 2.9 cm and 8.3 cm 2 might be suitable for patients at high risk of postoperative morbidity and mortality. Two-dimensional tumor size is an acceptable measurement method for estimating the risk of LN metastasis.

Background and Aims In individuals with either no or 1 to 2 nonadvanced adenomas, future risks of advanced colorectal neoplasia (AN) vary according to clinical risk factors. However, little is known about the association between the adenoma detection rate (ADR) and the risk for metachronous AN in patients with low-risk adenomas. Methods We identified 7171 participants with no or 1 to 2 nonadvanced adenomas at first-time screening colonoscopy. The risk of metachronous AN was investigated at surveillance colonoscopy, according to clinical characteristics and the ADR. Results In multivariate analysis the risk for metachronous AN was strongly associated with increasing age, male sex, increasing number of adenomas, and the ADR of the endoscopist. With the ADR modeled as a continuous variable, each 1.0% increase in the rate of ADR predicted a 3.0% decrease in the risk of metachronous AN (adjusted odds ratio [OR], .97; 95% confidence interval [CI], .95-.99). With the ADR modeled using a binary cut-off (32%), the risk of metachronous AN was reduced in patients of endoscopists with an ADR ≥32% (adjusted OR, .53; 95% CI, .35-.83). Moreover, the risk of metachronous AN was reduced (adjusted OR, .66; 95% CI, .46-.95) in patients of endoscopists with an ADR in the highest tertile, compared with patients of endoscopists with ADRs in the lowest tertile. The impact of ADR on metachronous AN was significant for patients with low-risk adenomas rather than patients with no adenoma. Conclusions In patients with low-risk adenomas, the ADR of the endoscopist was inversely associated with the risk of metachronous AN.