Multi-modality Medical Image Registration Based On Improved I-alpha Information With Gradient

In clinical diagnosis and treatment, it is usual to require Multi-modal lesion imaging for patient, In order to get the complementary, effective and comprehensive information, This requires information fusion of medical images. Will the information fusion of many images together, we can be in an image is reflected on a wide range of information. As a prerequisite for medical image data fusion, medical image registration has important clinical value, It is can be used for diagnosis and treatment, pathological changes can also be used for tracking and evaluation of treatment and other aspects.Medical image registration has become one of the focuses in the field of medical image processing early. Because of various objective factors (such as: different imaging equipment caused by the limitation image information) and subjective factors (such as:the interaction between the patient doctors), Medical image registration research has certain difficulty and complexity. At present many good algorithms have been proposed, but each medical image registration methods are only designed for a particular problem, with some limitations, and registration speed, accuracy and other aspects of registration can not simultaneously achieve the desired Effect.In this thesis, it is used the image registration algorithm based on gray level statistical model-registration mutual information model to do images registration for different modes medical images, Usually, based on the model of the image registration method is called maximum mutual information method, which uses information theory the mutual information as the image registration between the two to be Similarity measure does not require pre-segmentation of images, almost any image registration of different modes, and can get a good registration results.The Thesis introduces the development process of the mutual information for medical image registration, and analysis the local maximum problems existence of information medical image registration based on the traditional mutual, and analysis several improved mutual information image registration for the strengths and shortcomings, Using the SNI information get from the I-alpha information instead of the traditional mutual information to improve the speed of registration; According to the mutual information deficiency, combining the improvement of spatial information to depict the gradient information with the SNI information which only consider the gray-scale information to get the new measure. Use the new measure for the registration of different modes medical image. Results prove the convergence and registration precision are greatly improved, and solution the robustness problems of the traditional mutual information for medical image registration commendably.