Font Size: a A A

Study On The Medical Image Processing Technology Of Human Skeleton

Posted on:2018-03-18Degree:MasterType:Thesis
Country:ChinaCandidate:G ChenFull Text:PDF
GTID:2334330542490829Subject:Engineering
Abstract/Summary:PDF Full Text Request
Medical image processing technology is an important way to diagnose and treat diseases.Image segmentation technology has become the main part of medical image processing and analysis.Good image segmentation can bring convenience to the diagnosis and help doctors to diagnose correctly.In this paper the bone image segmentation method,in the study of spinal magnetic resonance imaging(MRI)and CT angiography(CTA)based on the characteristics of the skeleton,starting from the 2D image and 3D volume data segmentation method of skeleton images of different.First of all,according to the multi-scale normalized method proposed by Shi to MRI vertebrae segmentation easily appeared under segmentation and over segmentation or need too much manual adjustment of the parameters of the problem,we propose a method based on neighborhood information and chi square with Gauss weighted distance method and adaptive parameter adjustment of shared neighbor weighted.The pixels are arranged with the vector form of local neighborhood information,and then,considering the different effects of each pixel in the neighborhood of the central pixel,the weight should also be different,the Gauss function will be used chi square distance combinations,to avoid over segmentation or under segmentation.At last,by integrating the nearest neighbor weighted adaptive method,each pixel is automatically given a scale parameter to reduce the need to adjust the parameters.Secondly,after studying the segmentation method of MRI proposed by Zheng Qian,we introduce the local spatial neighborhood information of Gauss weighted,and construct the similarity matrix directly on the original image.At the same time,the NJW algorithm is used in the segmentation.When solving the eigenvector of the large weight matrix,it is complex and time consuming.It will restrict the actual application to the clinical.Based on the existing methods of graph-cut image segmentation,this paper analyzes how to construct a better network graph when the maximum flow minimum cut is used,and the key is to calculate the similarity between pixels.Considering the similarity algorithm proposed by Zheng Qian,the algorithm is more comprehensive.But in the graph cut method,the edge similarity algorithm doesn't make full use of the neighborhood information of pixels,so it needs to be improved.In this paper,the improved graph cut is used to replace the NJW algorithm in Zheng Qian's method,and the multi scale analysis method is applied to improve the efficiency of Zheng segmentation method,so as to enhance the clinical availability.Finally,based on the analysis of several existing medical image segmentation methods,a new method based on support vector machine is proposed.Before the reconstruction,the support vector machine is used to classify the data.Then,the voxel of the same kind of material is labeled the same,and then the gray value is transformed and arranged.The similar material transformation to the same gray interval,the interval is independent of each other,then its endpoint as piecewise point control opacity transfer function,to achieve the classification by Ray-casting(Direct Volume Rendering)method to complete three-dimensional vascular remodeling.
Keywords/Search Tags:medical image segmentation, local neighborhood information, multi-scale, CTA volume data
PDF Full Text Request
Related items