Title:AI Enabled High Frame Rate Portable Ultrasound Imaging Pipeline: Prototype Implementation with GPU Acceleration

Abstract:In this paper, we present a GPU-accelerated prototype implementation of a portable ultrasound imaging pipeline on an Nvidia CLARA AGX development kit. The raw data is acquired with nonsteered plane wave transmit using a programmable handheld open platform that supports 128-channel transmit and 64-channel receive. The received signals are transferred to the Nvidia CLARA AGX developer platform through a host system for accelerated imaging. GPU-accelerated implementation of the conventional delay and sum (DAS) beamformer along with two adaptive nonlinear beamformers and two Fourier-based techniques was performed. The feasibility of the complete pipeline and its imaging performance was evaluated with in-vitro phantom imaging experiments and the efficacy is demonstrated with preliminary in-vivo scans. The image quality quantified by the standard contrast and resolution metrics was comparable with that of the CPU implementation. The execution speed of the implemented beamformers was also investigated for different sizes of imaging grids and a significant speedup as high as 180 times that of the CPU implementation was observed. Since the proposed pipeline involves Nvidia CLARA AGX, there is always the potential for easy incorporation of online/active learning approaches.