Application Of Artificial Intelligence In Endoscopic Ultrasound Diagnosis Of Choledocholithiasis

Objective: Collected via endoscopic ultrasonography in the diagnosis of our department for images of the common bile duct calculi,by comparing the artificial intelligent,endoscopic ultrasonography physician joint endoscopic ultrasonography and artificial intelligent of common bile duct calculi diagnosed degree of sensitivity,specific and accurate indicators,such as research on artificial intelligence in the value of endoscopic ultrasonography in the diagnosis of common bile duct calculi.Methods: The case data and endoscopic ultrasound images of 964 patients diagnosed with choledocholithiasis by endoscopic ultrasonography from January 2018 to December2020 in the Center for Endoscopic Diagnosis and Treatment of Shengjing Hospital Affiliated to China Medical University were retrospectively collected,and the case data of 150 patients diagnosed with negative choledocholithiasis were also collected.The collected patient data were divided into training set(including 1967 images),verification set(including 100 images)and test set(including 150 cases of stone-positive patients and296 images).150 negative patients,359 images,a total of 300 cases,655 images).First by applying Easy DL for training,establish the model of A(test with and without stones)and model B(painted stone shape),secondly to debugging,the model has finished debugging model is then used to test set bravery manager endoscopic ultrasonography images and the common bile duct calculi endoscopic ultrasonography image recognition,finally calculated the sensitivity,specific degrees and accuracy of the diagnosis ability is evaluated.Results: For common bile duct calculi endoscopic ultrasonography images,treatment system of artificial intelligence diagnosis sensitivity,specific degrees and accuracy were75.81%,98.57% and 89.91%,respectively,are higher than experienced endoscopic ultrasonography physicians,sensitivity 94.42%,100.00% and 97.88% accuracy),when the two parallel diagnosis diagnosis way after test,artificial intelligent diagnosis system combined endoscopic ultrasonography physicians of common bile duct calculi diagnosed sensitivity was 97.67%,100%,the accuracy is 99.12%,the experience of medical ultrasonic endoscope has improved.Conclusion: As an auxiliary tool for clinical diagnosis and treatment system of artificial intelligence to assist the doctor endoscopic ultrasonography in the diagnosis of common bile duct calculi,thus improve the endoscopic ultrasonography diagnosis of common bile duct calculi identification and missed diagnosis,reduce lesions and artificial intelligent diagnosis system also improves the work efficiency of ultrasonic endoscope doctor,promote the popularization of endoscopic ultrasonography.