Ultrasound imaging is becoming increasingly essential in medicine, with software advances rapidly improving image quality and, with that, the clinical value. However, the deployment of these advanced methods necessitates substantial computing power, driving up system costs and limiting access to premium-quality ultrasound for many users.
To democratize premium ultrasound, this project proposes
to move all computational hardware off-device, onto the cloud. By doing so, users can leverage computing resources as needed, bypassing the purchase of expensive high-performance hardware upfront. This approach reduces the entry barrier to advanced ultrasound technology, ultimately expanding its accessibility.
We showcase the viability of the proposed architecture for latency-sensitive applications by developing a cloud-native ultrasound imaging demonstrator. The prototype features: (1) live streaming of ultrasound channel data to the cloud, (2) asynchronous cloud-based image formation, (3) image rendering to a web application, and (4) runtime parameters control.
To address the challenge of transmitting large data volumes to the cloud within critical latency constraints, we propose to add a local processor and a smart rate controller that ensure functioning even with interruptions of connection or with a time-varying network throughput.
The work conducted in this project initiates a new generation of ultrasound systems with the potential to disrupt the medical imaging industry.
