Quantitative study of cardiac motion estimation and abnormality classification in emission computed tomography

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Cardiovascular Imaging

Background:

• Quantitative cardiac motion description is crucial for detecting myocardial abnormalities in gated myocardial perfusion (GMP) emission computed tomography (ECT) images.
• Existing "optical flow" cardiac motion estimation (ME) techniques lack quantitative performance evaluation and application in abnormality quantification.

Purpose of the Study:

• To quantitatively evaluate cardiac motion estimation (ME) techniques for gated myocardial perfusion (GMP) emission computed tomography (ECT) images.
• To investigate the application of ME techniques in quantifying cardiac motion abnormalities.

Main Methods:

• Developed a 3D cardiac ME technique to generate a motion vector field (MVF) for voxel-by-voxel correspondence between GMP ECT images.
• Utilized weighted myocardial strain energy as a constraint to minimize intensity differences after warping.
• Optimized strain energy constraint weighting using noise-free and noisy phantom images for MVF accuracy and regional motion classification.

Main Results:

• Investigated ME technique convergence with varying initial estimates and cost functions, identifying dependence on initialization due to tangential motion.
• Optimized strain energy constraint weighting demonstrated effectiveness in both noise-free and noisy scenarios.
• Validated the optimized ME method's capability as a computer motion observer for differentiating normal and abnormal cardiac motion in GMP SPECT images.

Conclusions:

• The developed 3D cardiac ME technique provides a quantitative method for assessing cardiac motion from GMP ECT images.
• Optimized ME demonstrates potential as a computer-aided tool for detecting regional myocardial motion abnormalities.
• This approach enhances the diagnostic value of GMP ECT imaging for cardiovascular assessment.