Abstract #1204

Measuring Cardiac Strain in Duchenne Muscular Dystrophy with a Convolutional Neural Net Tag Tracking Method

Michael Loecher^1,2, Luigi E Perotti³, Patrick Magrath⁴, and Daniel B Ennis^1,2,5,6

¹Radiology, Stanford, Palo Alto, CA, United States, ²Radiology, Veterans Administration Health Care System, Palo Alto, CA, United States, ³Mechanical Engineering, University of Central Florida, Orlando, FL, United States, ⁴Radiology, University of California Los Angeles, Los Angeles, CA, United States, ⁵Cardiovascular Institute, Stanford, Palo Alto, CA, United States, ⁶Center for Artificial Intelligence in Medicine & Imaging, Stanford, Palo Alto, CA, United States

The objective of this work was to demonstrate the feasibility of using a convolutional neural net (CNN) based tag tracking algorithm for deriving strain measurements in grid tagged cardiac MR images. The method was tested in 23 subjects. When compared to commercial software the CNN-based method produces similar measurements for peak Ecc and shows lower strain in boys with DMD compared to healthy subjects [CNN = -0.15±0.03 vs -0.21±0.03] and [Conventional = -0.16±0.03 vs -0.21 ± 0.02] (p < .001). Peak Ecc was not significantly different within cohorts when compared between methods [DMD cohort: p=0.32, Healthy cohort: p=0.99]

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

Join Here