Microsoft and Cyted Use AI to Streamline Early Detection of Esophageal Cancer
-
Microsoft and Cyted collaborated to build AI models to detect esophageal cancer early by identifying Barrett's esophagus from capsule sponge samples. This could reduce pathologists' workload by up to 63%.
-
The AI models can accurately detect Barrett's using routine H&E staining, eliminating the need for additional costly TFF3 staining in most cases.
-
Two proposed workflows reduce pathologist workload by 52-63% while maintaining accuracy.
-
The code for the AI models is open source to enable others to adapt it to detect cancers in histopathology slides.
-
Cross-discipline collaboration between industry and academia can create innovations to improve early detection and patient outcomes.