Research And Implementation On The Fusion Of 3-D CT And Ultrasound Data In Cad System

In clinical diagnosis, doctors often use different medical imaging equipments at different time, different angles to complementary assist medical diagnosis. However, a variety of modalities as well as imaging conditions with the difference image resolution, contrast and spatial location caused by the different imaging mechanism lead to the need of fusion. The medical image fusion is what makes two images of different modality fused corresponding to their spatial location and anatomical structure by finding the spatial transform parameter between the images.The main task of fusion using 3-dimensional CT image and ultrasonic image includes the registration work and the presentation of the registed images of ultrasonic image obtained during the operation and the CT image before the operation. Currently, most of the efforts are devoted into the medical image fusion of CT images and MRI images. The fusion between CT and ultrasonic image is still in the initial stage. The ultrasound image has the instinct defect, that is, speckle noise caused by the inherent characteristics. So the mature approaches that widely used in the fusion of CT images and MRI images could not work in fusion work of CT images and ultrasound images. It request a novel 3-dimensional registration method based on the characteristics of CT imaging and ultrasonic imaging.The main work of this thesis is as follows:1. The preprocessing of CT and ultrasonic image before doing the actual registration work. We use erosion, dilation, opening image processing operations to reduce the noise in ultrasonic image and use gradient operation to generate the image outline.2. We developed a novel image fusion method to 3-dimensional CT images and ultrasound images based on the currently mature fusion methods. We first simulate the pseudo-ultrasound images from the CT images. Then we registed the pseudo-ultrasonic image and actual ultrasonic images3. We developed a ultrasound simulation method based on the ultrasound imaging principles using the acoustic impedance and reflection properties in the human organs.4. Two difference image similarity metrics---Mutual Information (MI) metric and Normal Vector Information (NVI) Metric are used in the registration work. And the feasibility of the metrics are evaluated using the developed system.5. We also developed an integrated presentation framework to display the result fusion images in 2-dimensinal and 3-dimensional ways. It can simultaneously display 2D and 3D images, CT and Ultrasound images and their fused images as well.6. The system can be used to lable, scale or rotate the ROIs in axial view, sagittal view and coronal view. An oblique presentation function is developed to enable the observation of the fused image from any perspectives. To meet the oblique view requirement of physicians, system can dynamicly display the plan normal to the operation path set by the users from beginning to target.