Improved Hessian Enhanced Adaptive Threshold Level Set For Hepatic Vessel Segmentation

The liver is the largest digestive and metabolic organ of the human body.In recent years,the world's liver diseases have occurred frequently,which seriously affects human health.The structure and morphology of the hepatic vessels can provide important information for the diagnosis,surgical planning,and outcome evaluation of the physician,and thus greatly facilitate the need to automate the segmentation of the liver vessels.In this paper,the segmentation method of liver vascular structure is studied,and the hepatic vascular structure with high precision is obtained.It is of great clinical significance to realize the threedimensional visualization of the results.The research content includes:(1)This paper used U-Net network for liver region segmentation and achieves high segmentation accuracy.The Sigmoid function is used to enhance the contrast of the image to provide high-quality input data for subsequent blood vessel enhancement and segmentation.(2)A blood vessel enhancement method based on Hessian filtering was implemented,which simultaneously enhanced the pixel points of the tube structure and the branch structure.Compared with the method of strengthening the vascular structure only as a tubular structure,the method has more advantages in enhancing the structure of the branch.(3)Combining the fuzzy membership function with the iterative threshold,adaptively adjusting the speed function,using different speed functions for different gray contrast areas to evolve,and completing liver blood vessel segmentation.After that,the 3D reconstruction of the blood vessel segmentation results and the establishment of a visual interface make the segmentation results more intuitive and have certain practical value.Compared with several classical methods in this field in recent years,this method can achieve higher segmentation accuracy and effectively reduce the over-segmentation phenomenon.Combining liver segmentation,blood vessel enhancement,and blood vessel segmentation methods realizes automatic segmentation of liver blood vessels,and also introduces overall performance evaluation indicators: accuracy rate,false detection rate,missed detection rate,and precision rate.This method is compared with this field in recent years.The comparison of the two new methods proves the effectiveness of the proposed method.