Study On Ultrasound Image Analysis And Navigation Methods For Interventional Diagnosis And Treatment Of Liver Diseases

With the development of the medical imaging, locating and tracking, and computing technologies, image-guided minimal invasive surgery has become one of the most important trends and methods on the lesions of clinical diagnosis and treatment. It is useful with real-time ultrasound image guided mainly because of its small surgical trauma, fast recovery and high surgical success rate. Unfortunately, since the ultrasound imaging mechanism may lead to the tunnel vision and low resolution, the ultrasonic interference causes a mass of speckle noises in ultrasound, the anatomical structures of ultrasound images such as organs and nidus during the ultrasound-guided minimal invasive surgery are not found as clear as that in CT and MRI. This can be improved by performing registration and fusion between 2D intraoperative ultrasound images and 3D preoperative CT images, to reduce the clinical surgery risks and operative time. The thesis, by focusing on the limitations of the current navigation system, studies several key issues: reconstruction of ultrasound volume and registration of ultrasound and CT. They can provide basis for the clinical application of navigation system and a simple, effective, minimal invasive therapy. Detailed work is introduced as following:(1) To improve the accuracy and complexity of current ultrasound simulation method, a novel method is developed for the simulation of an ultrasound image from CT slice. A multiscale method is used to simulate blood flow and to enhance tubular structures in the CT image. The acoustic response of common tissues are generated by weighted integration of neighboring pixels on the ultrasound propagation path in the CT image, from which parameters including attenuation, reflection, scattering and noise, are estimated simultaneously. The Kaiser Window function is then used to produce integration and radial blurring effects of multiple transducer elements to guarantee the simulated ultrasound image with high validity and calculation speed.(2) To solve edge blur from the traditional ultrasound volume reconstruction method, a novel global path matching method for 3D reconstruction of ultrasound images is proposed. Two operators are introduced to obtain the best contribution range for each pixel by calculating the invariant features of each pixel in the 2D ultrasound slices. The intensity of the filling vacant voxel can be obtained by weighted combination of the intensity distribution of pixels in the best contribution range. Three conditions are designed to calculate the weighting coefficient of the matching patch of the vacant voxel. While the matching patch is obtained by finding patches with the best similarity measure that defined by the conditions in the whole 3D volume data. Then, the intensity of the vacant voxel is replaced by voxels in the matching patch, for which the voxels in the patch are weighted contributed to the boundary voxels. Finally, the whole volume data is filled and ultrasound volume is obtained by iteratively repeating the above processing procedures.(3) Since the conventional similarity evaluation criteria is not suitable for the registration of ultrasound /CT, a novel method is developed for registration of ultrasound based on the ultrasound image simulation. An adaptive tensor-based principal component analysis method is used for ultrasound and CT image denoising. The ultrasonic simulation image based on CT slice is taken as the middle template. The correlation of linear combination is taken as the evaluation criteria for similarity. The registration of ultrasound/CT can be constrained with the improved Powell acceleration method.(4) Regarding the key issues of the navigation system, an ultrasound-guided liver minimal invasive surgery navigation system is constructed. Finite element model and image grid are introduced to solve the soft tissue deformation due to respiration and heart beating. After elastic registration between 2D intraoperative ultrasound image and 3D preoperative CT image, we can track and position the ultrasound probe with respect to the spatial position of the CT image using an electromagnetic tracking device, and then calculate the relationship between the ultrasound and CT. The fusion images can be displayed. The navigation system can assist clinical doctors to perform the liver minimally invasive surgery.