Noise reduction in estimating cardiac deformation from marker tracks

Area of Science:

• Biomedical Engineering
• Medical Imaging
• Computational Biology

Background:

• Cardiac wall deformation is crucial for assessing heart function.
• Current imaging techniques for measuring deformation are limited by pixel resolution and marker size.
• Accurate spatial measurement of markers is essential for precise deformation analysis.

Purpose of the Study:

• To improve the accuracy of cardiac wall deformation measurements using digital imaging.
• To investigate the effectiveness of Singular Value Decomposition (SVD) filtering for enhancing spatial accuracy.
• To validate SVD filtering in both computer simulations and open-chest animal experiments.

Main Methods:

• Utilized Singular Value Decomposition (SVD) filtering, assuming marker displacements are related within a continuum.
• Performed computer simulations with multiple markers across video frames.
• Conducted open-chest animal experiments using markers on the heart surface.
• Analyzed 256x256 video image sequences with precisely sized markers.

Main Results:

• SVD filtering improved marker position measurement accuracy from 0.14 to 0.045 (SD) pixels in simulations.
• In animal experiments, SVD filtering clarified deformation patterns without suppressing relevant high-frequency components.
• Marker position resolution improved from 0.1 to 0.03 (SD) pixels (6 microns) in a detailed experimental setup.
• Strain was determined with an accuracy of 0.002 over 30 pixels (6 mm).

Conclusions:

• SVD filtering significantly enhances the spatial accuracy of marker position measurements in cardiac deformation analysis.
• The method improves the clarity of deformation patterns in biological tissues.
• SVD filtering offers a robust approach for precise, high-resolution cardiac mechanics assessment.