An Ultrasonic Imaging SystemBbased on a New SAFT Approach and a GPU Beamformer

Abstract

The design of newer ultrasonic imaging systems attempts to obtain low-cost, small-sized devices with reduced power consumption that are capable of reaching high frame rates with high image quality. In this regard, synthetic aperture techniques have been very useful. They reduce hardware requirements and accelerate information capture. However, the beamforming process is still very slow, limiting the overall speed of the system. Recently, general-purpose computing on graphics processing unit techniques have been proposed as a way to accelerate image composition. They provide excellent computing power with which a very large volume of data can easily and quickly be processed. This paper describes a new system architecture that merges both principles. Thus, using a minimum-redundancy synthetic aperture technique to acquire the signals (2R-SAFT), and a graphics processing unit as a beamformer, we have developed a new scanner with full dynamic focusing, both on emission and reception, that attains real-time imaging with very few resources.

Software-based high-level synthesis design of FPGA beamformers for synthetic aperture imaging

“…With this input, the kernel would seek to form an intermediate image frame from the data of each transmission event. In line with the mathematical principles adopted in the GPU-based beamformer reported earlier [21], the computing process involves two operations: 1) conversion of each channel-domain rF data frame to an analytic signal form; and 2) derivation of image pixel values based on the analytic data samples using a two-way delay-and-sum procedure. subsequently, a compounded image (referred to as the sa image hereafter), is formed by coherently summing the intermediate image frames obtained from a group of transmission events.…”

“…For this task, we have assigned each work item to compute one pixel value in the intermediate image, and it involves four specific steps. First, the twoway focusing delays for all the array channels are individually calculated using well-established geometric formulas [21], taking into account the physical propagation time from the point source to the pixel position and then back to each array element. next, for each array channel, the two analytic data values at rF sampling points closest to the calculated focusing delay of that channel are fetched from the global memory buffer.…”

“…as a proof of concept, emphasis is placed on realizing the sa beamforming paradigm that has been applauded for its potential in achieving high imaging quality with two-way focusing at all pixel positions and whose GPU implementation has recently been realized [21], [22]. In approaching this task, our goal has been to achieve real-time and power-efficient execution of sa beamforming operations on FPGa that support high-level synthesis.…”

Riding the Plane Wave: Considerations for In Vivo Study Designs Employing High Frame Rate Ultrasound

“…Second, high-throughput computing hardware such as graphical processing units have greatly matured [27]. These parallel processing devices have served well to achieve real-time execution of HiFRUS-related computation tasks, such as pixel-by-pixel beamforming [28][29][30], Doppler processing [31][32][33], and post hoc filtering [34].…”

Cheetah: A Library for Parallel Ultrasound Beamforming in Multi-Core Systems

“…Commands for easily defining specific SAFT sequences of emission/reception are given. Anyway, the library comes with some predefined techniques, like 2R-SAFT [6], and TFM [9].  2D/3D imaging.…”