Research Of Registration Method And Accuracy Based On Medical CT Images

In recent years, with the development of medicine, computer and bio-engineering technology, medical imaging supplies multi-modality medical images for clinical diagnosis. A fundamental problem in medical image analysis is the integration of information from multiple images of the same subject, acquired using the same or different modalities and possibly at different time. In order to fuse the information from different images, an essential problem, which should be solved firstly, is to align one image to the other images. Medical image registration plays an important role in disease diagnosis, surgical navigation, monitoring the progression of disease, visualization of neurological surgery, as well as evaluation of the outcome of orthodontics, orthognathic surgery, bone grafting and joint repair.Medical image registration has become one of the hottest topics of the field of medical image research because of its important clinical value in recent years and various registration methods have been proposed. However, there are no "Golden" methods to achieve image registration in all cases, no common methods to judge registration accuracy and no uniform accuracy criteria. Therefore, based on Mimics, this paper selected the pre-and post-orthodontic treatment skull CT data, discussed three registration methods and their accuracy, as well as proposed corresponding methods of accuracy judgement, for the purpose of improving registration accuracy.In this paper, based on pre-and post-treatment skull CT data, the process and characteristics of "Image Registration" in Mimics have been researched, and on basis of MATLAB image processing toolbox, a new two-dimensional image processing modular, as well as an improvement method and an accuracy judgement menthod of "Image Registration" have been proposed. Then, by selecting pre-and post-treatment mandible CT data as research object, discussed the process of registration of3D models using "Point Registration" and "STL Registration" and analyzed their respective characteristics. On basis of Visual C++6.0, visualization software development tools and Magics, STL data processing software, two accuracy judgement methods, which can be available for "Point Registration" and "STL Registration", have been proposed, being named for accuracy judgement methods based on the average of shortest mutual distance and based on the average of overlap area. The scope and characteristics of these three judgement methods are discussed, with some corresponding application examples.These three accuracy judgement methods proposed in this paper would provide us a new approach to formulate a common accuracy judgement criteria in the future. Meanwhile, by using them for the registration of medical CT images, we could judge their relative accuracies, select an acceptable registration model with the highest relative accuracy as research objects of other aspects, which would be helpful for deep research on the mechanism of other aspects, and then improve the level of medical diagnosis to a certain extent.