| The use of computer processing and analysis of human CT images in mo dern medicine has become an important research direction and has extraordinary important clinical practical value.Human CT scans are rich in information,such as complex soft tissues,bones,blood vessels,and multiple organs.For computer-aided diagnosis,segmenting a region of interest(ROI)in a CT image is an essential prerequisite.Since the active contour segmentation method of medical CT images has the problems of being sensitive to the initial contour as well as the manual segmentation is time-consuming and laborious,it is urgent to find an automatic CT image method that can replace the manual segmentation.Aiming at the problem of being sensitive to the initial contour,time-consuming and labourious manual segmentation in the active contour segment ation method of medical CT images,this paper focuses on CT image data of human organs such as the brain,liver,lungs and vertebrae.An automatic Contour Segmentation of CT Images,which based on Pixels and Convolutional Neural Networks is proposed.In this method,superpixel meshing of CT images is firstly performed by superpixel segmentation.Secondly,superpixel classification is pe rformed through an improved convolutional neural network to determine the edge superpixels,and the seed points of the edge superpixels are extracted to form the initial contour.Lastly,based on the initial contour,the accurate segmentation of human organs is achieved by calculating the minimum integrated energy function proposed in this paper.Aiming at the algorithm in this paper,fast and accurate automatic segment ation of human CT images is achieved through simulation experiments on public data sets.In the Jaccard,Dice and CCR indexes of human organ segmentation,the algorithm proposed in this paper has high segmentation accuracy and obtains ideal segmentation results,which provides a theoretical basis and new solutions for clinical CT image lesion diagnosis. |
PDF Full Text Request