Comprehensive Summary
The present study by Abe et al. explores the proficiency of the Computer-Assisted Detection EYE system (CAD-EYE) prototype in the detection of esophageal squamous cell carcinoma (ESCC) and gastric neoplasm (GN). The CAD-EYE system is an artificially intelligent (AI) technology that provides clear, real-time images on an endoscopic screen, which the company Fujifilm further developed into a highly efficient detector for ESCC and GN. The study used data sets of video frames which were procured using three imaging modes: white light imaging (WLI), blue light imaging (BLI), and linked color imaging (LCI). Over 600 ESCC datasets were analyzed by the CAD-EYE system per imaging mode, and this was then compared to the “gold-standard” of lesion detection by well-experienced endoscopists. Over 800 GN datasets were analyzed for the WLI and BLI imaging modes. Resultant data from CAD-EYE analysis noted high specificity and sensitivity percentages with an average of over 90 percent in both metrics for both ESCC and GN detection, displaying highly promising applications.
Outcomes and Implications
The use of multimodal imaging to accurately analyze lesions in the body using AI has a high potential for bringing AI analyses into more center space. Abe et al. presents a strong framework for how CAD-EYE imaging could be implemented into more clinical settings for the safe and surefire diagnosis of upper gastrointestinal conditions. However, the study is limited due to the analysis of video frames versus actual live video analysis; it does not seem as good of a replacement to real-time endoscopy and analysis by an experienced endoscopist as it needs further training on differing anatomy, lighting conditions, and motion blur from movement through the body. Therefore, further testing needs to be done with the CAD-EYE system to fully test its capabilities prior to consideration for its use in real clinical settings.