An Efficient Parallel Architecture for Sub Band Processor in Broad Band MVDR Beamformer for Medical Ultrasound Applications

Abstract

Minimum Variance Distortion less Response (MVDR) algorithm is the key adaptive beamforming algorithm in high resolution Medical Ultrasound Imaging systems. MVDR Beamformer is a frequency domain beamformer, and each frequency sub band need to be processed to generate covariance matrix and the same is computationally costly. Though many VLSI architectures have been proposed for MVDR Beamformer, currently no Complex Correlation matrix generation architecture for frequency sub band has been devised which supports real time implementation of broad band MVDR Beamformer for Medical Ultrasound Imaging. In this paper we show real time MVDR sub array, sub band processor realization for an array with number of elements N=32 and sub array length L=8. To satisfy the narrow band condition of the array we employed 1K point FFT, with 12.5 % overlap. The entire design of MVDR Sub band processor is tested on the Field Programmable Gate Array (FPGA) platform. The implementation results show that the through put of the developed architecture is 600K Complex Correlation matrix generation per second which is equivalent to 65.57 fps to image a 20cm depth target at a Pulse Repetition Frequency (PRF) of 4 KHz, 600 Field of View and 10 lateral resolution and makes the MVDR a viable solution for practical systems.