(Poster) Using Fully Synthetic Training Data to Automate Clinical CT-ACR Phantom Analysis
Morgan Daly, John Hoffman,
We aim to illustrate the feasibility of using a convolutional neural network (CNN) trained solely on simulated input data to perform meaningful clinical medical physics tasks. We demonstrate this by automating detection and alignment of the HU module of the CT-ACR phantom (Gammex 464).
RSNA Annual Meeting 2023