Enhancement of Ultrafast Ultrasound Images: a Performance Comparison Between CNN Trained with RF or IQ Images

Ultrafast ultrasound imaging achieves high frame rates but results in lower image quality compared to focused acquisitions. Convolutional neural networks (CNNs) have emerged as effective tools for reducing diffraction artifacts and enhancing the quality of ultrafast ultrasound acquisitions. This study proposes a novel architecture specifically tailored for in-phase and quadrature (IQ) images. Our objective is to compare the impact of training CNNs with radio frequency (RF) and IQ images to discern the most effective approach for enhancing the quality of single plane wave (PW) acquisitions. Despite the networks trained with RF and IQ images yielding visually comparable results, the CNN trained with RF images outperforms the CNNs trained with IQ images in terms of peak signal-to-noise ratio and structural similarity index measure. However, analysis of in vitro phantoms indicates that training with IQ images produces superior quality speckle patterns.