Research Of Liver Segmentation From Abdomen CT Image Sequence And 3D Reconstruction

Liver diseases are common diseases that threaten human health worldwide,which have various kinds and develop very rapid.Otherwise many liver diseases are not easy to cure and recover.When we do the liver segmentation,the traditional Manually segmentation method depends much on the doctors’ work experience and prior knowledge so that there is difference between conditions during surgery and preoperative analysis,which will probably lead to the low success rate of the surgery.If we can segmant the liver accurately,reconstruct it,extract sagittal plane and coronal plane and calculate its volume from CT series images using computer image segmentation and 3D reconstruction technology so that the doctors can accurately master the data before the surgery,the success rate of surgery will be improved to some extent.So,Combining image processing,3D reconstruction techniques in the computer science field and anatomy,diagnostics in the medical field to complete liver CT sequence image processing and 3D reconstruction and then completing computer-assisted treatment of liver disease has become a hot topic at home and abroad.This article based on the characteristics of liver CT image sequence does the research of liver segmentation and three-dimensional reconstruction method of abdominal CT sequence.1.Completes the rough segmentation by region growing algorithm,histogram thresholding algorithm and the largest connected component algorithm.Region growing algorithm can roughly determine the location of the liver,which is possible to avoid the mis-segmentation of other organs in subsequent processing.The histogram threshold method can determine the gray value range of the liver and then further accurately segment.Maximum connected component algorithm can extract rough liver images.The segmentation process of rough segmentation uses OpenMP for parallel acceleration,and the execution efficiency is improved.2.Using GrowCut algorithm for optimal segmentation:some of the results of the above three-step rough segmentation still have defects.We can use the GrowCut algorithm that automatically generates seed markup templates for further optimizations.The morphological dilation and erosion operations can be used to extract the seed mark template of the contour generation GrowCut algorithm,and then it is used as an input to the GrowCut algorithm to perform the optimal segmentation.3.The ray casting algorithm is used to perform three-dimensional reconstruction of theliver based on the image sequences segmented in the first few steps.CUDA is used to speed up the reconstruction of the raycast volume,and the arithmetic of the matrix and vector are written as the inline device function,which will improve rendering efficiency.Finally,the sagittal and coronal planes of the liver are extracted and segmented for in-depth observation,diagnosis and treatment by clinicians and experts.