My talk at the NVIDIA GTC 2019 conference summarized most of our research on ultrasound imaging:
We discuss novel approaches to image acquisition, processing, and visualization that have the potential to radically change clinical practice and transform the ultrasound probe into an ever-more-indispensable point-of-care tool. In this session, we look beyond conventional image processing and show how more original applications of recent GPU-Based technologies like deep learning enable real breakthroughs in both diagnostics and interventions. We’ll talk about revisiting the full imaging pipeline with innovative real-time solutions including image enhancement, anatomy classification, or even 3D reconstruction of 2D clips without any tracking and deformable image registration. We’ll also introduce automated calibration algorithms and camera-like auto-focus based on real-time target detection, which simplifies ultrasound use to make it more accessible. Our talk will include live demos of algorithms within our framework, as well as its integration in NVIDIA’s CLARA platform.