The Segmentation Of Medical Ultrasound Images And Spatial Localization Of The Lesion Center

Surgical navigation system is a combination of modern engineering technology and medical technology. It can assist doctors to complete surgical procedures with advantages of reducing the difficulty of operations, improving the accuracy of surgery, decreasing surgical trauma and accelerating recovery etc. Now it has become the hotspot of both areas of computer technology and medical technology. Meanwhile, ultrasound imaging has been widely used in clinical diagnosis and treatment since its characteristics of real-time, non-destructive, inexpensive and flexible. Therefore, research on key technologies of ultrasound image-guided surgical navigation system such as image segmentation, image calibration,3D display and spatial orientation has been conducted in this paper.First of all, an optimization method of active contour models driven by region-scalable fitting and local Bhattacharyya distance energies (RSFB model, for short) with a double-well potential function and an automatic RSFB model with its initial contour determined by local image entropy are proposed on the basis of comparative studies of different image segmentation methods. Experimental results show that they can significantly reduce the running time of the algorithm and achieve automatic image segmentation respectively, thus having the potential to be applied in the ultrasound image-guided surgical navigation system.Second, a novel calibration approach for ultrasound images based on optical positioning is designed in this thesis. It can accomplish the calibration without introducing additional calibration template. Experimental results show that the calibration error is up to0.65mm when calibration images are more than20. This novel calibration approach works with high precision, good adaptability, strong flexibility and operational simplicity, making it a better choice for the ultrasound image-guided surgical navigation system.At last, a spatial localization method of the lesion center is presented to fulfill3D transformation and display of lesion contours in2D ultrasound images as well as the localization of the lesion center. Experimental results show that the deviations of the center location and the radius are within2mm and1mm respectively. The accuracy meets the requirements of the ultrasound image-guided surgical navigation system.