Precision medicine is gradually becoming an indispensable part of diagn osis,analysis,treatment and evaluation,especially for patients with liver dis ease.Accurate analysis of medical images can enable patients to obtain accurate diagnosis and shorten treatment time.An automatic segmentation method of liver CT image based on the Mask R-CNN algorithm provides an important automatic rec o gnition function for liver region detection.For the calculation of massive live r image samples,the algorithm framework of this paper adopts PyTorch and uses t he Cuda module to call the GPU instruction set to perform Effective parallel com puting.The results of the test image overlap and ROC accuracy show that the pre diction accuracy of this model reaches 92%,which can achieve the automatic reco gnition of the liver region and improve the liver resection rate,which provides effective auxiliary analysis for clinical medicine.