| Ultrasound-based elasticity imaging methods has developed rapidly in recently 20 years. As a new method of medical ultrasonic imaging, it can be easily to monitor the degree of hardenability of soft tissue tumor and bring the broad application and development prospect. However, because of the large-scale data and complexity of algorithm, it need long time to get a frame image. In order to improve the imaging speed and meet the real-time demand of elasticity imaging system, parallel computing technology based on GPU(graphics processing unit) has been concerned widely.The main research content of this paper is supersonic shear imaging. It is analyzed from the basic principle of elasticity imaging, the finite element model simulation analysis and GPU parallel algorithm. Details are as follows.In terms of imaging principle, firstly, this paper introduces the basic principle of ultrafast shear imaging and expounds the key steps in signal processing, including displacement and shear wave velocity estimation. Then the signal processing of the imaging system is improved by combining with the characteristics of the echo signal data.In terms of finite element method, this paper simulates the spatial distribution of acoustic radiation by Field II acoustic Field simulation toolkit. Then the tissue state of motion is simulated by using the ABAQUS software. Through the results of simulation, image quality can be evaluated.In terms of parallel algorithm design, this paper designs their algorithm based on CUDA(Compute Unified Device Architecture) and builds the framework of the GPU parallel computing according to the data structure and integrates the kernel functions. For the important processing tasks, this paper implements and optimizes their parallel algorithms. In the last, GPU parallel processing speed is tested by using experimental data. The results show that GPU can provide good speedup and ensure the real-time requirement in ultrasound imaging without losing the accuracy. |
PDF Full Text Request