Research On Multimodal Medical Image Fusion And Its Application

In recent years, due to the increasing demand for clinical application, the study multimodal medical image fusion attracted more attention. Physicians by identifying a large number of multi-modality medical images(such as CT, MRI) to diagnose the patient’s condition to determine the lesion, which requires a lot of time and experience. The contrast and complementary information on the existence of multi-mode medical image fusion into a picture, we can help physicians to better identify the characteristics are not easily observed, reducing the rate of wrong diagnosis and surgical error rate. Fused image but also through the subsequent processing, such as three-dimensional modeling, etc., for clinical or medical simulation teaching.Image processing has many advantages with other fusion method based on wavelet does not have, it’s multi-resolution analysis features, and not to increase data redundancy and have directionality, making wavelet analysis is very conducive to the integration of the image, has become a focus of research in the field. Therefore, this technique using wavelet transform fusion algorithm for multimodal medical images were basic research and implementation.Well localized in time domain and frequency domain and multi-resolution wavelet has, making wavelet analysis applications in image fusion has become a mainstream technology. Therefore, this technique using wavelet transform fusion algorithm for multimodal medical images were basic research and implementation.In the wavelet transform, the image will be broken down into two different low-frequency and high-frequency channels. For low-frequency channel, the paper verified by experiments, took the adaptive selection method based on local energy for coefficient selection, this method is suitable for handling low-frequency information, because most of the low-frequency information is centralized image energy. For the high-frequency channel is taken based on the human visual system SUSAN corner point extraction algorithm to the coefficient selection, as the image representing the high frequency channel is horizontal, vertical and diagonal edge information, using the SUSAN operator can effectively select edge higher response factor, but also be able to get better fused image. Finally, the wavelet coefficients obtained by performing inverse wavelet transform, fused image is obtained. On fusion algorithm proposed in this paper to experiment, from a subjective and objective analysis of both the experimental results show that the proposed algorithm can get better in the image fusion, can meet the accuracy requirements of medical images.On fusion algorithm proposed in this paper to experiment, from a subjective and objective analysis of both the experimental results show that the proposed algorithm can get better in the image fusion, can meet the accuracy requirements of medical images.Finally, the design and implementation of a system for brain surgery applications. The system consists of two three-dimensional model and image processing module, which has a three-dimensional model of model building and cutting module, the image processing is divided into segmentation, registration, fusion of three parts. Integration module which is a fusion of medical images based on the text given in the general framework, developed based on Matlab and C + + development tools, and the use of the text proposed algorithm implementation.