A Machine Learning Approach to Beamforming

Unmet Need
Given their ability to aid in real-time navigation of our complex anatomy, ultrasound and photoacoustic images are popular options for image guidance during surgery. However, noise in these images complicate surgical procedures, particularly noise caused by reflection and reverberation artifacts from highly echogenic structures, which often appear as true signals. If unidentifiable as noise rather than true signals, these artifacts cause irreversible damage to surrounding structures such as nerves and major blood vessels, leading to complications such as paralysis and patient death. Currently, none of the widely utilized ultrasound and photoacoustic image formation techniques account for these problematic reflection artifacts, which can arise from multiple factors that are intimately linked, yet difficult to model based on acoustic propagation theories alone.
 

Technology Overview
The technology takes a revolutionary machine learning-based approach to the formation of ultrasound and photoacoustic images. The approach is extendable to multiple applications where beamforming is required (e.g. radar, seismology).
 

Stage of Development
Developed method for removing artifacts in imaging by leveraging state-of-the-art machine learning techniques.