Design And Implementation Of Liver CT Image Automatic Segmentation System Based On Improved U-net

Liver cancer is one of the most common malignant tumors in China,and the most effective treatment is surgical treatment.CT is a common method for the examination of liver lesions.With the development of computer application,computer-aided liver surgery planning system can help surgeons to provide reasonable operation plan for patients before performing liver surgery,reduce the risk of operation and improve the success rate of operation.Liver image segmentation is a very important step in liver surgery planning,which can provide quantitative analysis for subsequent liver treatment.At present,manual segmentation and semi-automatic segmentation are mostly used in clinic,but the efficiency and accuracy of segmentation have not been further improved.Therefore,in order to reduce the workload of doctors and improve the efficiency and accuracy of segmentation,this paper proposes a medical image segmentation algorithm based on probability graph iteration based on deep learning,and combines this algorithm with software engineering to design and develop a system that can realize fast and automatic segmentation of liver CT images.The work of this paper mainly includes the following aspects:Firstly,aiming at the disadvantage of the loss of down-sampling information in U-net network,this paper proposes a new probability graph iterative network structure(Iterative-net).The specific method is to input the output of the network to the shallow layer of the network for iteration to compensate the information lost by down-sampling.The improved algorithm can effectively overcome the problem of information loss caused by pooling layer during network down-sampling in U-net.Secondly,the proposed algorithm is verified by experiments on the common data set Sliver07.First of all,the experimental data are enhanced by rotation and elastic deformation,and then all the data are normalized.After the completion of image preprocessing,the algorithm is trained,verified and tested.The experimental results show that the algorithm in this paper has good segmentation performance.Finally,this paper designs and develops the liver CT image automatic segmentation system for clinical application.Through the design of each functional module of the system,the functions of medical image reading and display,image preprocessing,image segmentation and image three-dimensional reconstruction are realized.At the same time,a comprehensive test of the system is done.It is proved that the system has a certain value in clinical application.