A Reseach To Liver Image Intelligent Segmentation Technique

Computer-assisted segmentation technology has increasingly played a prominent role in the medical research.Liver image segmentation as one of image processing technology in medical research and application contribute to liver medicine research,pathological analysis,preoperative planning,and evaluation.Therefore,the development of accurate,intelligent and efficient segmentation algorithm has been a common goal for medical image segmentation researchers.Blood flow,liver contraction and relaxation lead to a large number of noises in abdominal images,which brings a series of undesirable phenomena,including local gradient maximum area,artifacts and weak boundary.Those make a great difficulty to liver segmentation.In this regard,this paper presents a solution,mainly in the following areas.Firstly,we adopt an improved bilateral filtering algorithm,which can adaptively filter the abdominal CT images.Secondly,according to the multi-threshold maximum interclass variance algorithm based median,we extract the approximate mark of the liver.Finally,markers are used for morphological manipulation,and the watershed segmentation algorithm is used to perform accurate liver segmentation.By the presented methods,the liver segmentation accuracy has been significantly improved,and the segmentation process doesn't need the user intervention,greatly reducing the requirements of the user.This paper analyzes the effects of different noise reduction and segmentation methods,and several DICOM datasets are employed for proving the robustness of the presented method.After successfully dividing the liver,the liver vessels are extracted by ITKSNAP,which further demonstrates the integrity of the liver segmentation.