A graph theoretic approach for computing 3D+time biventricular cardiac strain from tagged MRI data.

Area of Science:

• Cardiovascular Imaging
• Biomedical Engineering
• Medical Physics

Background:

• Tagged magnetic resonance imaging (tMRI) is crucial for assessing regional heart function.
• Existing methods for computing cardiac deformation and strain from tMRI have limitations.
• Accurate 3D biventricular myocardial strain analysis is essential for understanding cardiac mechanics.

Purpose of the Study:

• To present a new, automated method for measuring 3D plus time biventricular myocardial strain from tMRI data.
• To improve the accuracy and reduce user intervention in cardiac strain analysis.
• To validate the proposed method using in-vivo human cardiac imaging studies.

Main Methods:

• Utilized a Gabor filter bank for tag point extraction from tMRI images.
• Employed a point classification with graph cuts (PCGC) algorithm and motion compensation for tag point classification.
• Computed 3D biventricular deformation and strain using a finite difference method on classified tag points.

Main Results:

• The proposed PCGC method achieved accurate 3D cardiac strain mapping with minimal manual correction.
• Automated processing required 44 minutes after initial contouring for a typical cardiac imaging study.
• Results were comparable to manual methods, demonstrating the technique's efficacy across various pathologies.

Conclusions:

• The developed method provides an accurate and automated approach for 3D plus time biventricular myocardial strain analysis from tMRI.
• This technique significantly reduces user intervention, making cardiac strain assessment more efficient.
• The findings support the clinical utility of this novel method for evaluating cardiac mechanical function.