US 12,266,444 B2
Synthetically trained neural network for MRI tag tracking
Michael Loecher, Santa Monica, CA (US); and Daniel B. Ennis, Palo Alto, CA (US)
Assigned to The Board of Trustees of the Leland Stanford Junior University, Stanford, CA (US); and The United States Government As Represented By The Department Of Veteran Affairs, Washington, DC (US)
Filed by The Board of Trustees of the Leland Stanford Junior University, Stanford, CA (US); and The United States Government As Represented By The Department Of Veteran Affairs, Washington, DC (US)
Filed on Jan. 19, 2021, as Appl. No. 17/152,302.
Claims priority of provisional application 62/963,123, filed on Jan. 19, 2020.
Claims priority of provisional application 62/963,124, filed on Jan. 19, 2020.
Prior Publication US 2021/0219862 A1, Jul. 22, 2021
Int. Cl. A61B 5/00 (2006.01); A61B 5/055 (2006.01); A61B 5/11 (2006.01); A61B 5/318 (2021.01); G06N 3/08 (2023.01); G06V 10/44 (2022.01); G06V 10/764 (2022.01); G06V 10/82 (2022.01); G16H 30/40 (2018.01)
CPC G16H 30/40 (2018.01) [A61B 5/0044 (2013.01); A61B 5/055 (2013.01); A61B 5/1128 (2013.01); A61B 5/318 (2021.01); A61B 5/7267 (2013.01); A61B 5/7289 (2013.01); A61B 5/748 (2013.01); G06N 3/08 (2013.01); G06V 10/454 (2022.01); G06V 10/764 (2022.01); G06V 10/82 (2022.01); A61B 2576/023 (2013.01)] 18 Claims
OG exemplary drawing
 
1. A method for magnetic resonance imaging (MRI) tag tracking, the method comprising:
synthetically generating synthetic tagged image data from natural images using programmed tag motion and a full Bloch simulation, wherein the synthetic tagged image data comprises dynamic training images and tag motion paths;
providing the synthetic tagged image data to a convolutional neural network (CNN), wherein the CNN is configured to generate grid tag motion paths indicative of cardiac motion;
acquiring MRI images using a tagged imaging method; and
determining, based on the grid tag motion paths indicative of cardiac motion, a path of tag lines through a cardiac cycle from the MRI images,
wherein generating the synthetic tagged image data comprises: applying periodic motion fields to the natural images to produce the dynamic training images, and generating tagged images from the dynamic training images using the full Bloch simulation, wherein the synthetic tagged image data are based on the tagged images.