Deep Learning for Cardiac Ultrasound Analysis

Partnered with french (Bordeaux University Hospital) and US (HSS New York) hospitals to create deep learning models for ultrasound image analysis.

Installed our own servers inside the hospital in order to keep the data in situ and maximize security.

Developed an annotation platform for doctors to label the cardiac ultrasound videos

Created 2 algorithms: one providing acquisition guidance, the second one deriving critical measures (ejection fraction, myocardium thickness) for cardiac diagnoses.

Integrated the models into ultrasound probes (Clarius) for testing purposes