Segmentation Algorithm Of Spine CT Image Based On Hidden Markov Random Field

The development of modern medical science greatly benefited from the progress of medical imaging technology,Medical image is divided into two parts of the imaging technique and image processing technology,Medical image processing,including image segmentation,image registration,image reconstruction,motion tracking,etc.Image segmentation is the basis of the subsequent processing,the result can be used for tissues and organs measurement,the structure reconstruction.Clinically,CT imaging is used in the bone tissue with higher density.However,because of the fuzzy effect of CT image itself,it brings no small challenge to computer automatic segmentation.In this paper,a segmentation method of spine CT image based on Hidden Markov random field is proposed.In order to verify the effectiveness of the proposed algorithm for the segmentation of spine CT images,the original image preprocessing,by highlighting edge vertebral bodies,reduce noise,to increase the contrast CT image of vertebral body more clearly.On the basis of preprocessing,the image using hidden markov random field model,The MLL model and Gauss mixture model are used in the labeling field and the characteristic field respectively.Then we use the maximum posterior criterion to present the energy function and transform the image segmentation problem into a mathematical problem.Finally,image segmentation using expectation maximization algorithm.Considering that the algorithm is easy to fall into local extremum,the global optimal solution is obtained by simulated annealing algorithm.But the processing speed of simulated annealing algorithm is not very ideal,so the idea of adding vibration points and condition iteration algorithm,On the one hand,remove the stable pixels before each iteration to reduce the number of pixels to be traversed perround.On the other hand,the conditional iteration algorithm is used to replace the intermediate iteration process of simulated annealing algorithm.In this paper,135 spine CT images are segmented on the MATLAB platform,and the segmentation results of the algorithm and the contrast algorithm are compared with the gold standard respectively to verify the accuracy of the algorithm.Then the paper compares the algorithm with the processing time of the algorithm,and obtains the advantage of the algorithm in the segmentation speed.