From June 2000 to October 2005 10, Professor Ding cooperated with Professor Aaron Fenster, a world-renowned authority on three-dimensional ultrasound and head of the Department of Biomedical Engineering at the University of Western Ontario. He studied the technology and application of three-dimensional ultrasound imaging, including needle detection, segmentation and tracking, prostate segmentation, radioactive particle segmentation and thrombus detection and segmentation. After more than four years of painstaking research, 2/kloc-0 articles have been published on the segmentation of three-dimensional ultrasound images, and the following main progress has been made:
1) A method of needle detection and segmentation based on projection is proposed. In order to reduce the influence of speckle noise, artifacts and high gray-scale points produced by other tissues, it is proposed to intercept three-dimensional images by using the rough information of needle position and direction provided by three-dimensional imaging system. At the same time, by selecting two mutually perpendicular projection directions and projecting by volume rendering technology, the segmentation of three-dimensional images is transformed into the detection of needles on two two-dimensional images, which greatly reduces the time required for segmentation. In order to ensure the correctness of needle detection in two-dimensional images, he also proposed a needle detection and tracking method based on real-time Hough transform. Related achievements were published in international authoritative journals such as Medical Physics (five-year impact factor IF=4.072), and an American patent was applied:
Needle detection, segmentation and tracking for 3D ultrasound-guided interventional procedures, A. Fenster, D. Downey, M. Ding, L. Gedi, American provisional patent application # 60/535,825: 2004 1 month 13. Ranked third.
2) A 3D prostate segmentation method based on slice is proposed. Firstly, the cardinal spline function is used to replace the traditional piecewise quadratic curve, which enhances the expression ability of complex prostate shape. Secondly, in order to solve the problem of segmentation error accumulation caused by the mapping and diffusion of prostate contour to adjacent frames, a series consisting of the distance from the contour boundary point to the intersection point with the rotation axis is established on the plane perpendicular to the rotation axis. The zero-order autoregressive model is used to represent the continuity limit, and the correct position of the contour boundary point is obtained. At the same time, an internal force is introduced through these points, which forces the contour to jump out of the local extreme points in the process of deformable segmentation search and obtain the correct prostate contour. Finally, the three-dimensional curved surface of prostate is generated from the contour edges formed on different sections. Related achievements have been published in international conferences such as Medical Physics and SPIE Medical Imaging.
3) A seed segmentation method based on three-cylinder projection is proposed. Firstly, according to the result of needle segmentation, the three-dimensional ultrasonic image containing seeds is intercepted. Secondly, high-brightness non-seed points, such as those formed by calcified tissue, are eliminated by the projection of three mutually perpendicular cylinders. Finally, the difference image is calculated along the rotating scanning direction of the ultrasonic image, and then the gray sum and direction of the possible straight line segments at each point are detected by using the three-dimensional straight line segment template. The point where the number of parallel straight line segments in the neighborhood exceeds a given value is the seed point. The research results were published at the 2005 SPIE Medical Imaging Conference held in San Diego, USA in February 2005, which aroused great interest from colleagues, including Mayo Clinic and Acory Company. June 5438+October 2005 10, I applied for an American patent:
Automatic particle localization and segmentation of intraoperative brachytherapy, M. Ding, L. Gedi, A. fenster, application number 60/647,420: 2005 1 28th. Ranked first.
Because of Professor Ding's outstanding work in the application of three-dimensional ultrasound, Professor Bir Bhanu of the University of California, Riverside invited him to write a chapter on "Visualization and Segmentation Technology of Three-dimensional Ultrasound Images" in his monograph "Computer Vision Beyond Visible Spectrum". Dr. Jasjit Suri, a world-renowned expert in medical image processing and a professor in case western reserve university, USA, also invited him to write a chapter on prostate ultrasound image segmentation based on deformable models in their monograph "Parametric and Geometric Deformable Models: Applications in Biomaterials and Medical Imaging" (Volume-II). Since 2002, he has been invited to give a speech at the annual SPIE International Conference on Medical Imaging. In 2003, he gave a speech at the 7th MICCAI International Conference held in Montreal, Canada. In September, 2005, he was invited to give a special report at the 27th EMBC International Conference of IEEE held in Shanghai on "Segmentation and Classification of Anatomy and Pathology from Medical Images", and served as the chairman of the small seminar and the chairman of the "Image Guided Surgery II" branch.
Professor Ding has rich experience in scientific research and product development. 1990, has been engaged in scientific research and product development related to image processing. With Shaanxi Color Picture Tube General Factory and Foshan Kangsida Hydraulic Co., Ltd., we have successively studied products such as on-line automatic detection of mixed color of fluorescent powder in color picture tubes, on-line quality detection and control of magnetic wall and floor tiles, and carried out image segmentation, target recognition and on-line with the support of projects such as National Natural Science Foundation, National High-tech 863, National Science, Technology and Industry Committee for National Defense Fund and Doctoral Program Fund of Ministry of Education. Especially since 2000, I have studied in Robarts Institute of Canada, a world-famous medical imaging research institution and the main birthplace of three-dimensional ultrasonic imaging technology in the world, for five years, and have mastered the world-leading three-dimensional ultrasonic imaging and application technology. At the same time, Dr. Ding actively promoted the industrialization of technology. In 2007, he initiated the establishment of Huazhong University of Science and Technology-Dizheng Yahe Medical Imaging Joint Laboratory, and served as the director of the joint laboratory.