Project Overview and Rationale for an Administrative Supplement The hemispheric array ultrasound breast imaging system currently being developed at the University of Rochester with the support of NIH grant R01 EB009692 is a unique facility comprised of a data-acquisition apparatus and a hybrid data-acquisition and high-performance computer network. The system leverages significant advances in ultrasound transducer arrays, front-end electronics, digital technology, and theoretical breakthroughs in inverse scattering to provide speckle-free, high- resolution, quantitative images of intrinsic tissue characteristics, i.e., sound speed and attenuation slope. The system is designed to acquire data during a two -second interval by using 10,240 parallel channels for transmission and reception and to image the entire breast volume with isotropic point resolution as good as the lateral resolution of x-ray mammography by use of a novel reconstruction algorithm. Using non-ionizing ultrasound, this system permits risk-free examination of the breast for cancer detection and overcomes limitations of x-ray mammography such as low resolution of contrast in dense breast, distortion and discomfort resulting from compression-induced deformation of anatomy, and poor imaging of breasts with implants. Demonstration of success would ultimately change the way screening for breast cancer is performed and significantly improve detection, diagnosis, and monitoring of response to treatment of breast cancer. An important theoretical development ? a major breakthrough ? has occurred since the original grant was awarded. This breakthrough allows our unique reconstruction algorithm to reconstruct different subvolumes independently, i.e., in parallel. At the same time that our algorithm was being extended, graphical processing units (GPUs) were evolving into powerful parallel computational engines. These GPUs allow small high-performance computer systems (HPCs) to perform massively parallel computations that are ideally suited for implementation of our parallelized reconstruction algorithm. Incorporation of GPUs in the computing nodes of our system endows the hybrid computer network with a computational capability comparable to that of the large, federally-supported national supercomputer facilities that were originally intended to be used for reconstructions. A GPU-based HPC network coupled to our data-acquisition system will be able to produce breast images in 10 ?15 minutes rather than the days that would be required on a supercomputer using our original single-volume method. Reconstruction times measured in minutes rather than days would have an enormous impact on the clinical utility of our imaging instrument because data could be acquired and the reconstructed volumes of the breast viewed in the course of a single visit. An ultrasound imaging system capable of producing images with 100-micron resolution in minutes would provide an improved and efficient way to screen for breast cancer and also to diagnose other breast disease.
