Research On CT-MRI Image Information Fusion Algorithm Of Upper Limb

Fracture incidence rate ranks fourth in the world common diseases,in which limb fractures accounted for more than 70% of the total cases.At present,the degree of soft tissue injury and repair after fracture surgery is a blind spot in the research field,It mainly relies on the patient's subjective experience and the physician's clinical experience when diagnosing,There is not a complete evaluation system.In this paper,we explore a CT-MRI image information fusion algorithm based on the human upper limb structure in order to observe the postoperative bone healing and soft tissue growth around the bone,it is a great significance in clinical diagnosis and rehabilitation evaluation of human limb fractures.Region-of-interest segmentation is the basis of image registration and fusion.Compared with the edge detection results of Sobel,Laplacian and Canny,there are obvious multiple edges and fragmentation edges,and the boundary integrity is poor,although some regional segmentation methods can guarantee the edge integrity,they have over segmentation and segmentation error problem such as regional seed growth and flood fill algorithm,Using the man-machine interaction method of watershed algorithm to detect,this algorithm restrains the generation of multiple edges,solves the problem of over-segmentation effectively and completes the upper region CT-MRI image segmentation and boundary extraction of soft tissue accurately.Then through the expansion and erosion of the segmentation boundary,the seed-point is selected to realize the intelligent segmentation of adjacent tomographic images.According to the anatomic features of the upper region,a rigid-flexible registration scheme was proposed.The improved SIFT model which add a backtracking function based on the forward matching is used to rigidly match the skeletal area of the upper limb,which not only improves the precision of the registration point,but also accurate to sub pixel level registration.Introduced by Freeman chain code to smooth the contour line,the tissue region is matched elastically.The contour line is closer to subjective perception because the edge zigzag was smoothed.And then,the tissue boundary of the CT image was fitted to the MRI image by down-sampling and interpolation technique,which realize the pixel level registration of the soft tissue region.Finally,the spatial structure fusion method is proposed,based on traditional interpolation method,layer fusion method and wavelet fusion method.In this method,we carry out the intelligent fusion with different weights in different regions,using the information of skeletal area from CT image and the information of soft tissue area from MRI image.The spatial structure fusion method has its own advantages after quantitative evaluated by information entropy,image clarity,cross entropy and other indicators.For the twelve samples of different sex,age and body weight,the method has the best effect on 85% samples after the subjective evaluated by professional physicians.The algorithm effects.The CT-MRI image information fusion algorithms of upper limb was proposed in this paper.It can not only fully reflect human skeletal information,but also provide the clear information of soft tissue.That provides the basis for clinical diagnosis and rehabilitation evaluation of limb fractures.