Validation of a Software-Defined 100-Gb/s RDMA Streaming Architecture for Ultrafast Optoacoustic and Ultrasound Imaging

TL;DR

This paper addresses the bottlenecks in optoacoustic and ultrasound imaging by proposing ListenToLight (LtL), a software-defined 256-channel architecture that unifies OA and US data acquisition with continuous raw-data streaming. The approach combines wideband analog front-ends, a tightly integrated Zynq MPSoC (APU+PL), and a 100 GbE RDMA backend to achieve sustained throughput up to 95.6 Gb/s, validated on a 16-channel prototype while demonstrating 256-channel scalability. Core innovations include a buffer-free data path, JESD204B-based high-density data transfer, deterministic clocking with SYSREF, and software-controlled sequencing and RDMA data movement. Phantom experiments confirm end-to-end functionality for both US and OA, and the framework is shown to outperform many state-of-the-art open and commercial platforms in bandwidth, openness, and dual-modality capability, with a clear path toward larger channel counts and integrated acquisition modules.

Abstract

Optoacoustic (OA) imaging has emerged as a powerful investigation tool, with demonstrated applicability in oncology, neuroscience, and cardiovascular biology. However, its clinical translation is limited with the existing OA systems, which often rely on bulky and expensive acquisition hardware mainly optimized for pulse-echo ultrasound (US) imaging. Despite the fact that OA imaging has different requirements for receive bandwidths and timing synchronization with external laser sources, there is a strong need for unified OA-US imaging platforms, as pulse-echo US remains the standard tool for visualizing soft tissues. To address these challenges, we propose a new data acquisition architecture for ultrafast OA and US imaging that fully covers the requirements for large channel counts, wide bandwidth, and software-defined operation. LtL combines state-of-the-art wideband analog front-ends, a Zynq UltraScale+ MPSoC integrating FPGA fabric with an Application Processing Unit, and a 100 GbE Remote Direct Memory Access (RDMA) backend enabling raw-data streaming at up to 95.6 Gb/s. The architecture avoids local buffers followed by burst transfers, which commonly constrain sustainable frame rate and recording intervals, thus achieving true continuous and sustained streaming of raw data. We validate the core elements of the LtL architecture using a 16-channel demonstration system built from commercial evaluation boards. We further verify the signal chain for up to 256-channel scalability, confirming the wide bandwidth capabilities to support state-of-the-art data transmission speeds.

Figures (8)

• Figure 1: Radar plot comparing key performance metrics of LtL against state-of-the-art (SoA) hybrid OA--US imaging platforms (see Table \ref{['tab:platform-comparison']}). Axes are normalized to $[0,1]$. Modality: $US=0.5 OA+US=1$ . Pulser: $s_{\mathrm{TX}}=\tfrac{1}{3}s_{\mathrm{levels}}+\tfrac{1}{3}s_{V_{\mathrm{pp}}}+\tfrac{1}{3}s_{f_{\max}}$. AFE/ADC: $s_{\mathrm{AFE}}=\tfrac{1}{2}s_{N_{\mathrm{bits}}}+\tfrac{1}{2}s_{f_s}$. Buffer*: max on-system memory; for host-buffered systems: equivalent to max. on-system buffer.
• Figure 2: Top: Overview of the LtL system architecture. Bottom: 16-channel demonstration implementation (red boxes indicate the differences between the 256-channel system concept and the 16-channel validation).
• Figure 3: JESD implementation (adapted from bhattacharjeeListenToJESD204BLightweightOpenSource2025).
• Figure 4: Overview of 100G RDMA interface architecture
• Figure 5: RX characterization.