A PC-based fully-programmable medical ultrasound imaging system using a graphics processing unit

In this paper, a PC-based fully-programmable medical ultrasound imaging system is presented where a high-performance graphics processing unit (GPU) is utilized to perform entire ultrasound processing. In the proposed architecture, ultrasound signal and image processing algorithms were divided into four modules and efficiently implemented on the NVIDA's Computer Unified Device Architecture (CUDA) platform (GeForce GTX285, NVIDA, Santa Clara, CA, USA). To evaluate the real-time performance of the proposed architecture, 64-channel, 128-scanline pre-beamformed radio-frequency (RF) data were captured with a commercial ultrasound machine (G40, Siemens Healthcare, Mountain View, CA, USA). The execution time was measured by examining the time stamp produced by a CIDA timer. For generating a 800×600 ultrasound B-mode image, it takes 18.32 ms for single-beam based ultrasound processing with the acquisition time of 19.7 ms. These results indicate that the GPU-based fully-programmable system architecture can support real-time ultrasound signal and image processing.

[1]  Yongmin Kim,et al.  A Fully Programmable Computing Architecture for Medical Ultrasound Machines , 2010, IEEE Transactions on Information Technology in Biomedicine.

[2]  Li-wen Chang,et al.  Graphics processing unit-based high-frame-rate color doppler ultrasound processing , 2009, IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control.

[3]  Tai-Kyong Song,et al.  A fractional filter-based beamformer architecture using postfiltering approach to minimize hardware complexity , 2007, IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control.

[4]  G York,et al.  Ultrasound processing and computing: review and future directions. , 1999, Annual review of biomedical engineering.

[5]  Tetsuya Hayashi,et al.  A single mediaprocessor-based programmable ultrasound system , 2003, IEEE Transactions on Information Technology in Biomedicine.