MR guided catheterization requires both fast imaging and fast reconstruction techniques for interactive imaging. Recent deep learning methods outperformed classical iterative reconstructions with shorter reconstruction times. We propose a low latency framework relying on deep artefact suppression using a 2D residual U-Net with convolutional long short term memory layers trained on multiple orientations. The framework was demonstrated to reconstruct an interactively acquired bSSFP tiny golden angle radial sequence for catheter guidance. The proposed approach enabled real-time imaging (latency/network time=39/19ms) in 3 catheterized patients with promising image quality and reconstruction times.
This abstract and the presentation materials are available to members only; a login is required.