With the advent of AI technology, many trials for medical purpose have started. Regarding the field of AI medical imaging, two major ways are CADe for the detection of the lesions, and CADx for the diagnosis. Moreover, some case reports are available in CAP (Computer-aid Prediction) especially in the field of optical fundus and mammary gland.
There are two issues for AI research in the field of biliary tract diseases coming up as follows: 1. Each modality is more likely to have lesion-specific detectability. e.g. CT-negative bile duct stones are often detectable using extracorporeal and endoscopic ultrasound. 2. Pre-treatment diagnosis is often made by the results from multiple modalities.
To solve these problems, the authors recommend the Japan Biliary Association firstly collect the big data including the images and texts from all kinds of biliary related modalities, and start with CADe for the lesion-specific detectability in the dedicated modalities and CADx using AI algorithm generated from those data. GAN and cluster technology would be of great help for rare biliary diseases.
The authors expect the future AI research in the biliary tract diseases fruitful.
View full abstract