Vector Quantization of RF Channel Data for Ultrasound Imaging

Abstract

Quantization is a necessary step in a digital beamformer that should be implemented efficiently when designing an ultrasound imaging system. Several factors should be considered, including the accuracy of both amplitude and phase in the digital representation, the quantization efficiency, and the complexity of the quantization scheme. While uniform scalar quantization (SQ) is currently used in most ultrasound imaging systems, this study used simulations and experimental data to explore vector quantization (VQ) based on the signals received by adjacent elements of the transducer array generally being strongly correlated. The signal-to-quantization-noise ratio, focusing quality, and accuracy in phase aberration correction and blood velocity estimation were assessed. The results show that using VQ instead of conventional SQ reduces the quantization noise by 10 dB with minimal degradation in the focusing quality. Similarly, the performance of correlation-based phase aberration correction was not affected at this level of quantization. Nevertheless, velocity estimation accuracy was found to be more susceptible to VQ noise when the reduction in the number of bits is limited to a factor of two. One benefit of using a more efficient quantizer is the reduced data size, which directly affects the data transfer rate and is an important design consideration for a digital imaging system.