Motion estimation in cardiac fluorescence imaging with scale-space landmarks and optical flow: a comparative study

Area of Science:

• Cardiovascular Physiology
• Biomedical Imaging
• Computational Biology

Background:

• Motion artifacts, primarily pixel misalignment from cardiac contraction, significantly degrade optical mapping recordings.
• Accurate action potential analysis in cardiac studies is hindered by these motion-induced artifacts.

Purpose of the Study:

• To develop and evaluate methods for identifying landmarks and tracking cardiac tissue motion in optical mapping.
• To establish a foundation for landmark-based image registration to correct motion artifacts in cardiac fluorescence videos.

Main Methods:

• Comparison of Scale-Invariant Feature Transform (SIFT) keypoints and global optical flow (OF) for motion estimation on low-contrast cardiac images.
• Application of both SIFT and OF to track pixel motion in cardiac optical mapping videos with simulated motion.

Main Results:

• SIFT demonstrated superior performance over OF for pixel motion tracking in low-contrast, low-resolution cardiac optical mapping images.
• Landmark-based image registration using SIFT resulted in improved action potential recovery and action potential duration calculations.

Conclusions:

• SIFT is an effective method for landmark detection and motion tracking in challenging cardiac optical mapping images.
• SIFT-based registration offers a promising solution for mitigating motion artifacts and enhancing the accuracy of cardiac electrophysiological studies.