We introduce a novel approach to fitting parameters from DCE MRI using an unsupervised neural network. The network is trained on in vivo data, with no ground truth, and is able to predicts DCE model parameters directly from the obtained MRI images. In simulations, our method outperformed the ordinary least squares fit approach in that it is more accurate and precise. In vivo, it produced substantially less noisy parameter maps than the current practise least-squares fit.
This abstract and the presentation materials are available to members only; a login is required.