Research On Medical Ultrasound Imaging Techniques Using Parallel Computing

Diagnostic ultrasound is an important part of modern medical-imaging modality. It is routinely used in all clinical medicine specialties because it is easy to use, safe, noninvasive, and relatively inexpensive. The medical ultrasound imaging is promoted to a higher level by the rapid development of the computer, electronics, information technology. Hence, diagnostic ultrasound has made great progress. The US extended-filed-of-view imaging technology overcomes the disadvantage of the limited image FOV of a real-time ultrasound scanner, provides a more easily interpretable image and provides a panoramic image containing larger organs and their surroundings to expand diagnostic capabilities.Three-dimensional ultrasound imaging technology breakthroughs some of the limitations of the conventional two-dimensional ultrasound images in clinical application, and many clinical benefits that have been identified include the ability to obtain views not possible with 2D imaging, the increased understanding of complex geometric structures, and the ability of accurately measure distances and volumes.In ultrasound imaging, its salient features are large volumes of data to be processed and processing and display the results in real-time. Moreover, the image processing algorithms are computationally intensive. Hence, the real-time performance of image processing systems has become increasingly demanding, and requires computer with high computing performance.In recent years, with the increasing performance of microprocessors and network technology continues to evolve, the use of relatively inexpensive computers with high-speed LAN interconnection to build high-performance parallel computer cluster computing systems is becoming increasingly popular. Compared with traditional supercomputers, parallel cluster computing system is simple deployment and cost-effective and good scalability to meet the different sizes of large computational problems.In order to achieve real-time imaging performance, we have designed a parallel image registration algorithm using SIFT for EFOV sonography and a parallel reconstruction algorithm for three-dimensional ultrasound on cluster of PCs.The experimental results show that the image registration algorithm using SIFT for EFOV sonography can produce EFOV image with high accuracy in comparison with a traditional method, when the rotations among the captured images are relatively large. Meanwhile, our parallel algorithm overcomes the high complexity of the SIFT algorithm and the intensive computation of EFOV sonography, and speeds up the generation of panorama with the speedup over 7 on a cluster of 5 PCs with a Dual-Core CPU; the parallel reconstruction algorithm for three-dimensional ultrasound has been speedup more than five times on the cluster; moreover, the two parallel algorithms have good scalability, when adding more PCs into the cluster, the algorithms can get more speedup without any modifications to the structure of the algorithm.