Using NVIDIA Holoscan to Remove Artifacts in Ultrasound Imaging

Опубликовано 28 ноября 2021, 14:00

Ultrasound Color Doppler Imaging is a non-invasive way to visualize blood flow in the body and detect blockages in arteries or blood clots in veins. However, during a scan, blood flow can exceed the rate at which instruments can accurately gauge blood flow speeds; and aliasing artifacts can obscure visualization.

The LITMUS group at the University of Waterloo found a way to remove these aliasing artifacts from visualizations by using over 1000 images to train a U-Net convolutional neural network to detect them. They then used the NVIDIA Clara Developer Kit, Clara Holoscan SDK, CUDA, and NVIDIA TensorRT plus an ultrasound front end platform from us4us to not only remove the aliasing artifacts but also increase the frames per second by 12 fold from 2 frames per second to 30 frames per second.

Learn more: nvda.ws/30QsEN9

Follow us:
X: nvda.ws/4hi2sxB
LinkedIn: nvda.ws/4eSlE3B
Subscribe to the NVIDIA AI Healthcare Newsletter: nvda.ws/4eS1F54