1. Open-Source AI for More Reliable Measurement of Artery Narrowing
Quantitative coronary angiography (QCA) is a computer technique that measures very precisely how narrowed a coronary artery is, providing more objective and consistent results, especially in scientific studies. This widely used technique still requires numerous manual corrections. The team of physicians and scientists from EPFL and CHUV developed AngioPy Segmentation, a deep-learning tool for rapid and accurate segmentation.
A custom annotation tool was first created by the image analysis hub, allowing Thabo Mahendiran, a physician and researcher at CHUV and EPFL, to annotate the data. A deep learning model was then trained together with Edward Andò, head of the Image Analysis Hub at the Center for Imaging with this dataset. Called AngioPy Segmentation, the tool automatically identifies the entire artery on the basis of a few points selected by a physician along the blood vessel by generating a mask that predicts which areas belong to the artery and which do not.
You are an EPFL scientist and you nee help with your images? Contact the hub for image analysis