Our Innovative Deep Learning Project for Automated Puncture in Ultrasound Imaging

August 2023 –

We are happy to share some news regarding our ongoing project addressing the localization of blood vessels in ultrasound images using deep learning. To enable automated puncture, a computer must be able to identify the blood vessels in the images taken by the ultrasound probe.
This skill is developed through a process called machine learning requiring a suitable dataset. A dataset was built by capturing ultrasound images of volunteers. This dataset was used to develop a machine-learning model capable of detecting superficial blood vessels in ultrasound images.
The position, size, and shape of the vessels are evaluated to determine which vessel is best suited for puncture. The current scientific work is supervised by the Institute of Medical Technology and Intelligent Systems at Hamburg University of Technology, which brings in profound expertise in image processing and artificial intelligence.