Research On Key Techniques Of Freehand Three Dimensional Ultrasound Based Surgery Navigation System

With the help of medical imaging techniques, space positioning techniques and computer graphics techniques surgery navigation systems could be helpful in surgery planning and surgery navigation. It can decrease the difficulties of surgeries, improve the accuracy and reduce the harm to normal tissues. It has been one of the fastest developed surgical techniques.Most of the available surgery navigation systems are achieved with the help of preoperative CT or MRI images. As they register the images with marked points on surface of the patients, registration errors are inevitable. If the tissues drifted during the operation it could not be reflected on these preoperational images. All these limitations has limited the navigation accuracy and hindered their application in chest and abdomen areas.Freehand three dimensional ultrasound based surgery navigation system use the three dimensional images of freehand ultrasound, which can be collected and updated during the operation. It performs over the available systems in real-time capability and accuracy, and is significant and promising in many clinical applications. As an emerging technology, it has to solve many technical problems. This thesis is dedicated to some of the key techniques in this system.The positions of each 2D ultrasound image must be known so as to get three dimensional ultrasound images. But the images and position information are not collected synchronically due to the limitations of the positioning systems and image capturing devices. Then we have to get the delay and figure out some algorithm to match each image with its position. In this thesis, we proposed an algorithm to get the delay and designed an algorithm of linear interpolation so as to match images to proper positions.To get the position relationship between the 2D ultrasound images, we must get the transformation matrix between the coordinate of the probe and the coordinate of the positioning systems. We implemented a calibration algorithm using the wall phantom. Firstly, we scan the phantom from a variety of angles and positions. Then, we use some a proposed algorithm to detect the position of the phantom in the images. In the end, Levenberg-Marquardt algorithm is used to solve the equations.To meet the demand for accuracy and speed, we proposed an adaptive square distance weighted reconstruction algorithm. It can adapt the size of the interpolation kernel according to ambient voxels. When the kernel expanded to a specific size, it becomes to pixel nearest algorithm which is less complex. To further contract time needed for reconstruction, we used parallel computing based on multi-core CPU and compiling optimization with Intel compiler in the implementation of the algorithm. Experiments demonstrated that these measures can reduce computational complexity without the loss of image quality.Sectional mode display is the most frequently used visualization technique for 3D ultrasound. Traditionally, we must first reconstruct the volume data and then use some method to extract the slice from the 3D data. This makes the reconstruction time consuming and with low accuracy. We developed a direct reconstruct algorithm for arbitrary slice reconstruction from irregularly arranged 2D images. With no need for volume data reconstruction and only one interpolation process, it can be faster and more accurate.For the tracking of the surgery devices, we must get its position in the coordinate of the positioning device. This is also a problem of calibration. The accuracy of the calibration determined the accuracy of the navigation system. In this thesis we put forward a method for the calibration. The key points'position in the coordinate of the positioning system can be obtained by moving the devices in some restrained ways.