High-Performance FPGA Architecture for Multi-line Beamforming in Ultrasound Applications

Nowadays, several applications are requested to process huge amount of data with tight temporal constraints to produce "real time" results. This is the case of ultrasound (US) tomographic (B-mode) images used in diagnostics. A B-mode image is typically composed by many tens of vertical lines. Each line is calculated by the so-called beamformer (BF) that combines the echoes received from tens of small piezoelectric transducers integrated in the US probe. The BF processes massive quantity of data (several Gb/s) and is implemented in fast digital devices like the FPGAs. In the standard approach, the B-mode image is formed on a line-by-line basis, i.e. one line per pulse repetition interval (PRI). This limits the achievable image frame rate (IFR) to few tens of Hz, which is not suitable to detail fast, important phenomena like the heart valve movement. In the so-called plane-wave approach, the data needed to produce all of the B-mode image lines are acquired in a single PRI. In this case, the IFR can be in the order of kHz, but a more complex BF is needed. In this paper, we present a Multi-Line BF architecture, capable of real-time processing tens of lines per PRI. The BF was implemented in the research ultrasound system ULA-OP 256. A sample B-mode sequence obtained at IFR = 600Hz is presented.