Geometric regularization for 2-D myocardial strain quantification in mice: an in-silico study

Area of Science:

• Cardiovascular Imaging
• Biomedical Engineering
• Computational Biology

Background:

• Myocardial strain quantification using 2-D speckle tracking in mice is challenging due to low frame rate-to-heart rate ratios.
• Significant decorrelation between frames necessitates robust regularization methods for accurate motion estimation.
• Geometric regularizers, enforcing specific motion field characteristics, offer potential but require comparative evaluation.

Purpose of the Study:

• To compare the performance of different geometric regularizers for myocardial motion and strain estimation in murine echocardiography.
• To evaluate regularizer effectiveness in simulated normal and infarcted heart models.
• To identify optimal regularization strategies for improving strain quantification accuracy.

Main Methods:

• Utilized simulated echocardiography datasets of murine hearts (normal and infarcted models).
• Applied and compared two geometric regularization methods: spatial curvature constraint and Gaussian convolution of lateral motion.
• Assessed myocardial strain estimation accuracy using root-mean-square (RMS) error for radial and circumferential strains.

Main Results:

• In normal models, spatial curvature regularization worsened radial strain (RMS error: 0.06 to 0.09) but improved circumferential strain (RMS error: 0.035 to 0.01).
• Gaussian convolution improved circumferential strain estimation (RMS error to 0.015) in normal models.
• In infarcted models, curvature regularization significantly improved circumferential strain (RMS error: 0.043 to 0.015), while Gaussian convolution improved it in remote regions (RMS error to 0.021).

Conclusions:

• Geometric regularization techniques offer significant improvements in myocardial strain quantification in mice, especially for circumferential strain.
• The choice of regularizer impacts strain estimation accuracy, with curvature and Gaussian methods showing differential benefits in normal and infarcted states.
• These findings aid in optimizing echocardiographic analysis for murine cardiovascular research.