Study On The Segmentation Method Of Liver And Its Tumor Based On CT Image

Liver diseases are a huge threat to human health.The clinical diagnosis and treatment of such diseases rely on computer-aided technology heavily.Therefore,how to obtain the location and size of the liver and liver tumors from CT images is an important prerequisite for liver disease diagnosis,surgical planning,functional evaluation,and treatment decision-making.However,relying on professionals to perform purely manual segmentation is time-consuming and laborious.In addition,because the liver and tumors exhibit the characteristics of blurred edges,low contrast,and uneven gray in CT images,it is also more difficult to auto-segment by computers.Therefore,according to the characteristics of liver and liver tumor in CT images,this study established a semi-automatic segmentation method to achieve segmentation efficiently and accurately.The main research contents are as follows:For the liver,this study uses an algorithm based on an improved level-set model for segmentation.The level-set model driven by the single information of the region or edge is difficult to complete the liver segmentation task.The level-set model introduces edge information on the basis of the regional level-set based on global and local information to construct a hybrid level-set model,and binarize the edge indicator function in the model,and use the area growing algorithm to extract the initial contour of the level-set evolution to complete the liver segmentation task.The algorithm was used to segment 40 sets of human abdominal CT image sequences(from Sliver07 and 3Dircadb),and good segmentation results were obtained.The 20 sets of MR image sequences in Sliver07 are segmented,and the segmentation effect is also good.In addition,this study uses bone information to construct liver segmentation constraints.The algorithm obtains the bone information through the threshold method firstly,then uses the ellipse fitting algorithm to find feature points,and finally uses the cubic interpolation algorithm to construct constraints.Using this algorithm to constrain some CT images with missing edges at the liver and muscle adhesions,effectively solving the over-segmentation problem caused by the liver and muscle adhesions during the liver segmentation process.For liver tumors,this study uses an improved fuzzy C-means clustering algorithm to complete semi-automatic segmentation.This method first locates liver tumors through human-computer interaction.Then use the gray distribution of the target area to directly obtain the cluster center and the membership degree matrix,classify the pixels according to the membership degree,and complete the liver tumor segmentation.The algorithm is used to segment the tumors in the 3Dircadb dataset,and the segmentation results with smaller error results were obtained.Finally,the proposed segmentation algorithm is used to segment other human organs and animal organs,and relatively ideal segmentation results are also obtained,which proves the robustness and applicability of this research algorithm.