Title: Sa2057 COMPARISON OF ARTIFICIAL INTELLIGENCE AND EXPERT ENDOSCOPIST TOWARD REAL-TIME ASSISTED DIAGNOSIS OF ESOPHAGEAL SQUAMOUS CELL CARCINOMA.
Abstract: Narrow-band imaging (NBI) is currently regarded as the standard modality for diagnosing esophageal squamous cell carcinoma (SCC). We developed a computerized image-analysis system for diagnosing esophageal SCC by NBI and estimated its performance with video images. Altogether, 7181 non-magnified (non-ME) and 7530 magnified (ME) endoscopic images of 1571 pathologically confirmed superficial esophageal SCCs and 564 non-ME and 2744 ME images from non-cancerous tissue or normal esophagus were used as training data. Five- to nine- second video clips from 144 patients were selected as validation data. These video images were diagnosed by the AI system and 13 board-certified specialists (experts). The diagnostic process was divided into two parts—the first was detection (identify suspicious lesions), and the other was characterization (differentiate cancer from non-cancer). The sensitivities, specificities, and accuracies for the detection of SCC were, respectively, 91%, 51%, and 63% for the AI system and 79%, 72%, and 75% for the experts. The sensitivity of the AI system was significantly higher than that of the experts, but its specificity was significantly lower. The sensitivities, specificities, and accuracy for the characterization of SCC were, respectively, 86%, 89%, and 88% for the AI system and 74%, 76%, and 75% for the experts. The receiver operating characteristics curve showed that the AI system had significantly better diagnostic performance than the experts. Our AI system showed significantly higher sensitivity for detecting SCC and higher accuracy for characterizing SCC from non-cancerous tissue than endoscopic experts.