Research And Application On Multimodality Medical Image Registration And Fusion

With the development of medical imaging technology, the multimodality medical images have been used widely in medicine. These images have their own advantages as well as disadvantages. Different modality medical images can provide complementary information for medical diagnosis. In order to use the information completely, these images must be registered and fused first. So the registration and fusion of multimodality medical images is helpful to improve the accuracy of clinical diagnoses and surgical therapies.This paper mainly studies on multimodality medical image registration and fusion, and uses these technologies to resolve registration and fusion of CT images and color slice images of "digital pig" data set, realizes the registration and fusion of the two different modality images in two-dimensional space, and further carries out segmentation and 3D reconstruction of the fused images for the ultimate goal of establishing 3D model of "digital pig". The main contents of this thesis as follows:1. According to the features of the above-mentioned two different modality images, color slice images are chosen as the reference images when the CT images are registered in the registration process, and select feature points base on interactive feature registration method, then the parameters of affine transformation are worked out by this feature extraction method, and finally the image registration is realized.2. In image fusion process, according to contrast pyramid method and IHS method, a new method is proposed based on a combination of the above two methods. Firstly, use contrast pyramid fusion method to deal with the I component images of color slice images and CT images, then do IHS fusion on the obtained images and color slice images to get the final images. Experimental results demonstrate the effectiveness of the method.3. The research situations and methods of medical image segmentation are reviewed, and a new segmentation method is proposed based on the GDI + and narrowband M-S model. Firstly, use GDI + segmentation technology to get the rough outline of tissues and organs from the fused images, then take the outline as initial zero level set curve, and with the narrowband M-S model, ultimately get a more accurate profile. Experimental results demonstrate the effectiveness and advantages of the proposed method.4. A new method is used on 3D reconstruction, which based on AMIRA and 3DSMAX. First, reconstruct the segmented organs data set and bounding box in AMIRA, then use AMIRA and 3DSMAX to mesh simplification and smoothing the initial 3D model, finally obtain the 3D model. Experimental results demonstrate the excellent performance of this method.