Real-time Artificial Intelligence for Detection of Upper Gastrointestinal Cancer by Endoscopy: a Multicentre, Case-Control, Diagnostic Study.
Huiyan Luo,Guoliang Xu,Chaofeng Li,Longjun He,Linna Luo,Zixian Wang,Bingzhong Jing,Yishu Deng,Ying Jin,Yin Li,Bin Li,Wencheng Tan,Caisheng He,Sharvesh Raj Seeruttun,Qiubao Wu,Jun Huang,De-wang Huang,Bin Chen,Shao-bin Lin,Qin-ming Chen,Chu-ming Yuan,Hai-xin Chen,Heng-ying Pu,Feng Zhou,Yun He,Rui-hua Xu
DOI: https://doi.org/10.1016/s1470-2045(19)30637-0
2019-01-01
Abstract:Background Upper gastrointestinal cancers (including oesophageal cancer and gastric cancer) are the most common cancers worldwide. Artificial intelligence platforms using deep learning algorithms have made remarkable progress in medical imaging but their application in upper gastrointestinal cancers has been limited. We aimed to develop and validate the Gastrointestinal Artificial Intelligence Diagnostic System (GRAIDS) for the diagnosis of upper gastrointestinal cancers through analysis of imaging data from clinical endoscopies. Methods This multicentre, case-control, diagnostic study was done in six hospitals of different tiers (ie, municipal, provincial, and national) in China. The images of consecutive participants, aged 18 years or older, who had not had a previous endoscopy were retrieved from all participating hospitals. All patients with upper gastrointestinal cancer lesions (including oesophageal cancer and gastric cancer) that were histologically proven malignancies were eligible for this study. Only images with standard white light were deemed eligible. The images from Sun Yat-sen University Cancer Center were randomly assigned (8:1:1) to the training and intrinsic verification datasets for developing GRAIDS, and the internal validation dataset for evaluating the performance of GRAIDS. Its diagnostic performance was evaluated using an internal and prospective validation set from Sun Yat-sen University Cancer Center (a national hospital) and additional external validation sets from five primary care hospitals. The performance of GRAIDS was also compared with endoscopists with three degrees of expertise: expert, competent, and trainee. The diagnostic accuracy, sensitivity, specificity, positive predictive value, and negative predictive value of GRAIDS and endoscopists for the identification of cancerous lesions were evaluated by calculating the 95% CIs using the Clopper-Pearson method. Findings 1 036 496 endoscopy images from 84 424 individuals were used to develop and test GRAIDS. The diagnostic accuracy in identifying upper gastrointestinal cancers was 0.955 (95% CI 0.952-0.957) in the internal validation set, 0.927 (0.925-0.929) in the prospective set, and ranged from 0.915 (0.913-0.917) to 0.977 (0.977-0.978) in the five external validation sets. GRAIDS achieved diagnostic sensitivity similar to that of the expert endoscopist (0.942 [95% CI 0.924-0.957] vs 0.945 [0.927-0.959]; p=0.692) and superior sensitivity compared with competent (0.858 [0.832-0.880], p<0.0001) and trainee (0.722 [0.691-0.752], p<0.0001) endoscopists. The positive predictive value was 0.814 (95% CI 0.788-0.838) for GRAIDS, 0.932 (0.913-0.948) for the expert endoscopist, 0.974 (0.960-0.984) for the competent endoscopist, and 0.824 (0.795-0.850) for the trainee endoscopist. The negative predictive value was 0.978 (95% CI 0.971-0.984) for GRAIDS, 0.980 (0.974-0.985) for the expert endoscopist, 0.951 (0.942-0.959) for the competent endoscopist, and 0.904 (0.893-0.916) for the trainee endoscopist. Interpretation GRAIDS achieved high diagnostic accuracy in detecting upper gastrointestinal cancers, with sensitivity similar to that of expert endoscopists and was superior to that of non-expert endoscopists. This system could assist community-based hospitals in improving their effectiveness in upper gastrointestinal cancer diagnoses. Copyright (C) 2019 Elsevier Ltd. All rights reserved.