A New Framework for Performing Cardiac Strain Analysis from Cine MRI Imaging in Mice

Area of Science:

• Cardiovascular Imaging
• Medical Artificial Intelligence
• Biomedical Engineering

Background:

• Cardiac magnetic resonance (MR) imaging is crucial for assessing cardiac function.
• Strain analysis provides comprehensive diastolic myocardial function assessment, often missed by standard cine imaging.
• Traditional tagged MR imaging for strain analysis is challenging in small animal models like mice due to heart size.

Purpose of the Study:

• To develop a novel deep learning approach for automated quantification of cardiac strain from cine MR images.
• To enable accurate assessment of myocardial function in small animal models.
• To validate the developed algorithm against established methods and demonstrate its utility in differentiating between healthy and diseased hearts.

Main Methods:

• A fully convolutional neural network (FCN) architecture was used for accurate localization of the left ventricular (LV) blood pool center-point.
• LV regions of interest (ROIs) were extracted for LV cavity and myocardium segmentation using another novel FCN.
• A Laplace-based approach tracked LV wall points between successive cine frames, followed by Lagrangian-based strain estimation.

Main Results:

• The automated system accurately quantified cardiac strain from cine MR images.
• Validation against tagged MR imaging showed no significant differences in strain data.
• The algorithm successfully identified strain differences between normal and diseased hearts.

Conclusions:

• A novel deep learning framework enables automated, accurate cardiac strain analysis from cine MR imaging in mice.
• This method overcomes the limitations of tagged imaging in small animal models.
• The automated system is a valuable tool for assessing myocardial function and disease in preclinical research.