Research On Medical Image Fusion And3D Reconstruction

Medical image fusion is a process. Firstly, a set of images are collected by the same ordifferent medical imaging apparatus generated under different environmental conditions forthe same target. Secondly, use the correlation image fusion techniques to extract themaximum degree of each image in the useful data and process related information. Finally,converged to generate a target image. Not only improves the fused image information of theimage content, and make it easier for the user to detect, identify, and can add more to identifythe target scene information. Medical image reconstruction theory is to reconstruct thethree-dimensional image of the target object by processing discrete3D data, while its field ofknowledge including computer science, graphics, image theory and relevant medicalknowledge, so it is a complex interdisciplinary. Medical image fusion and medical imagereconstruction technique is very effective for improving the accuracy of diagnosis.CBF (Cerebral Blood Flow) image contains a wealth of spectral information and thebrain CT (Computed Tomography) images with a high spatial resolution. If these two imagestogether, the fused image can be improved not only the spectral information may also makethe spatial resolution of the image is improved. For an objective evaluation to measure thequality of the fused image fusion algorithm is effective is critical, therefore thesis commonobjective evaluation were studied. Medical image fusion problem for color and grayscalemedical images, we propose a fusion algorithm based on fast HIS (Hue Intensity Saturation)Transform and WT (Wavelet Transform) combine. Fuse HIS strength component and grayimage by multi-scale wavelat transform, high-frequency part with the regional energyintegration rules, however, for the low-frequency portion with the weighted average rule. Themain idea of the algorithm is as follows:(1) to get I (intensity) component of CBFmulti-spectral by HIS transform the and then using wavelet decomposition to get highfrequency part and low frequency part of I component and panchromatic CT images.(2)According to the WT multi-scale features, high frequency part and low frequency part ofimage respectively apply regional energy fusion rule and weighted average fusion rule toproduce fused results.(3) To obtain final fused using fusion results raplace Inew of fast HIStransform.The algorithm compares the image of three experiments with other algorithms, theexperimental results show that the algorithm in mutual information (Mutual Information), structural similarity (Structual Similarity) and QFABother objective indicators significantlybetter than other algorithms, and by comparing the source image and fused image can befound in the algorithm to solve the problem of spectral distortion HIS transformation.Medical image reconstruction technique is mainly divided into surface renderingalgorithms and volume rendering algorithms in two ways. In this paper, the main principlemethod，the algorithm processes and ambiguity problem of MC (Marching Cubes) algorithmfor surface rendering are studied and analyzed. At the same time ambiguity problem is solvedto use the theory of hyperbolic asymptote. Three-dimensional images of the human brain inbones were reconstructed by experiments and the images reconstructed can be rotated,zoomed and draged. For volume rendering, the paper uses the ray-casting algorithm and studythe principle of algorithm processing steps as well as the synthesis of sequential samplingpoints. Three-dimensional images of the human brain volume rendering have been rebuildedby experiments. In addition, the same three-dimensional image can rotate, zoom and drag andso on.