Abstract #0784

An Analysis of DiffNet Reconstruction Performance in Healthy and Infarcted Cardiac Diffusion Tensor Images

Tyler E. Cork^1,2, Eric Aliotta³, Michael Loecher¹, Luigi E. Perotti⁴, and Daniel B. Ennis¹

¹Radiological Sciences Laboratory, Stanford University, Stanford, CA, United States, ²Bioengineering, Stanford University, Stanford, CA, United States, ³University of Virginia, Charlottesville, VA, United States, ⁴Radiological Sciences, University of California - Los Angeles, Los Angeles, CA, United States

Cardiac diffusion tensor imaging (cDTI) suffers from low signal-to-noise ratios, which results in tensor variability. In order to decrease tensor variability, the number of diffusion directions or number of averages must increase, consequently increasing the scan time. Recent implementations of artificial neural network (ANN) have proven that a non-linear mapping between diffusion signals and tensors is possible and can decrease tensor variability without increasing scan time. We implement an ANN tensor reconstruction for ex vivo porcine hearts to evaluate if a robust ANN diffusion tensor reconstruction is a beneficial technique to decrease tensor variability at no cost in scan time.

This abstract and the presentation materials are available to members only; a login is required.

Join Here