Cardiac motion estimation by joint alignment of tagged MRI sequences

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Computational Anatomy

Background:

• Cardiac displacement fields are crucial for understanding heart function.
• Existing image registration methods for tagged MRI have limitations in joint image alignment.
• Previous approaches used linear combinations of cost functions, limiting accuracy.

Purpose of the Study:

• To develop a novel metric for joint image alignment in tagged MRI using k-Nearest Neighbors Graphs (kNNG).
• To enhance the accuracy of cardiac displacement field recovery.
• To apply the method for studying myocardial infarction and its relation to cardiac strain.

Main Methods:

• Application of k-Nearest Neighbors Graphs (kNNG) estimators of α-entropy (H(α)).
• Measurement of joint similarity between cardiac MRI frames.
• Combination of information from different cardiac views into a unified metric.

Main Results:

• Significantly higher accuracy (p<0.05) compared to standard pairwise alignment (PA) in terms of mean positional error and variance.
• Demonstrated consistency between cardiac strain, infarction location, and coronary occlusion in patients.
• Validated the clinical utility of graph-based metric estimators in medical imaging.

Conclusions:

• kNNG estimators provide a more accurate method for joint image alignment in tagged MRI.
• The developed technique improves the analysis of cardiac mechanics and aids in diagnosing myocardial infarction.
• Graph-based metric estimators offer valuable solutions for practical medical imaging problems.