Deep Liver Lesion AI System: A Liver Lesion Diagnostic System Using Deep Learning in Multiphase CT

In the medical field, computer-aided diagnosis (CAD) systems are an important tool that can assist doctors in the diagnosis process. However, the powerful higher performance and accuracy models in CAD systems require high computation costs. Furthermore, collaboration between doctors is problematic when the CAD system is run on a single machine. To solve these problems, we developed a deep liver lesion AI system for a liver lesion diagnostic system based on a web foundation, in which core processing and data storage are implemented on the backend system. The core processing functions of the system consist of 5 modules: liver segmentation, liver lesion detection, liver lesion segmentation, liver lesion classification, and hepatocellular carcinoma early recurrent prediction. The visualization and interaction functions between the system and doctors are implemented on the web interface. With all processing functions implemented on the backend and the interface shown with a web interface, data sharing between doctors becomes easier than the standalone version.