MedPrax Report

CADDIE (Computer-Aided Detection for Endoscopy) is a cloud-based AI software developed by Odin Vision, designed to assist endoscopists in detecting and characterizing colorectal polyps during colonoscopy procedures. Leveraging NVIDIA's cloud infrastructure, CADDIE analyzes real-time video feeds from standard white-light endoscopes, providing visual markers to highlight potential polyps. It classifies polyps into adenomas and non-adenomas, aiding clinicians in making informed decisions about resection and histopathology. CADDIE also assesses bowel cleanliness by displaying the percentage of visible mucosa and identifies the appendiceal orifice to confirm cecal intubation. Its cloud-based nature allows for seamless software updates without the need for on-site interventions. CADDIE is compatible with existing endoscopy equipment and integrates into current clinical workflows with minimal training required.

The ASUS EndoAim AI Endoscopy System is a compact, real-time artificial intelligence solution designed to assist gastroenterologists in identifying and classifying polyps during colonoscopy procedures. Utilizing Intel Core processors and the OpenVINO toolkit, EndoAim processes 60 frames per second with a latency of less than 70 milliseconds, enabling rapid detection of polyps, including those that are small or difficult to detect. It highlights potential polyps on-screen and classifies them as adenoma or non-adenoma, providing real-time feedback to clinicians. Additionally, EndoAim offers a one-click size measurement feature, aiding in the assessment of polyp dimensions without the need for traditional visual estimation methods. EndoAim integrates seamlessly with existing colonoscopy equipment, requiring no additional hardware beyond a mini PC, and has been implemented in over 30 medical institutions in Taiwan.