Layer-selective deep representation to improve esophageal cancer classification.

Even though artificial intelligence and machine learning have demonstrated remarkable performances in medical image computing, their accountability and transparency level must be improved to transfer this success into clinical practice. The reliability of machine learning decisions must be explained and interpreted, especially for supporting the medical diagnosis. For this task, the deep learning techniques' black-box nature must somehow be lightened up to clarify its promising results. Hence, we aim to investigate the impact of the ResNet-50 deep convolutional design for Barrett's esophagus and adenocarcinoma classification. For such a task, and aiming at proposing a two-step learning technique, the output of each convolutional layer that composes the ResNet-50 architecture was trained and classified for further definition of layers that would provide more impact in the architecture. We showed that local information and high-dimensional features are essential to improve the classification for our task. Besides, we observed a significant improvement when the most discriminative layers expressed more impact in the training and classification of ResNet-50 for Barrett's esophagus and adenocarcinoma classification, demonstrating that both human knowledge and computational processing may influence the correct learning of such a problem.

Rio de Janeiro Global Consensus on Landmarks, Definitions, and Classifications in Barrett's Esophagus: World Endoscopy Organization Delphi Study.

BACKGROUND & AIMS: Despite the significant advances made in the diagnosis and treatment of Barrett's esophagus (BE), there is still a need for standardized definitions, appropriate recognition of endoscopic landmarks, and consistent use of classification systems. Current controversies in basic definitions of BE and the relative lack of anatomic knowledge are significant barriers to uniform documentation. We aimed to provide consensus-driven recommendations for uniform reporting and global application. METHODS: The World Endoscopy Organization Barrett's Esophagus Committee appointed leaders to develop an evidence-based Delphi study. A working group of 6 members identified and formulated 23 statements, and 30 internationally recognized experts from 18 countries participated in 3 rounds of voting. We defined consensus as agreement by ≥80% of experts for each statement and used the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) tool to assess the quality of evidence and the strength of recommendations. RESULTS: After 3 rounds of voting, experts achieved consensus on 6 endoscopic landmarks (palisade vessels, gastroesophageal junction, squamocolumnar junction, lesion location, extraluminal compressions, and quadrant orientation), 13 definitions (BE, hiatus hernia, squamous islands, columnar islands, Barrett's endoscopic therapy, endoscopic resection, endoscopic ablation, systematic inspection, complete eradication of intestinal metaplasia, complete eradication of dysplasia, residual disease, recurrent disease, and failure of endoscopic therapy), and 4 classification systems (Prague, Los Angeles, Paris, and Barrett's International NBI Group). In round 1, 18 statements (78%) reached consensus, with 12 (67%) receiving strong agreement from more than half of the experts. In round 2, 4 of the remaining statements (80%) reached consensus, with 1 statement receiving strong agreement from 50% of the experts. In the third round, a consensus was reached on the remaining statement. CONCLUSIONS: We developed evidence-based, consensus-driven statements on endoscopic landmarks, definitions, and classifications of BE. These recommendations may facilitate global uniform reporting in BE.