Feature-based MRI data fusion for cardiac arrhythmia studies.

Area of Science:

• Biomedical Engineering
• Cardiovascular Research
• Medical Imaging

Background:

• Current methods for studying cardiac arrhythmias (electrical/optical recordings, limited transmural data, models) lack 3D detail and completeness.
• Arrhythmias occur in 3D myocardial tissue, necessitating advanced imaging for comprehensive analysis.

Purpose of the Study:

• To demonstrate that combining Magnetic Resonance Imaging (MRI)-based Current Density Imaging (CDI) and Diffusion Tensor Imaging (DTI) can enhance the assessment of cardiac arrhythmia dynamics.
• To utilize feature-based data fusion to quantify and improve complementary information from electrical current distribution and structural heart properties.

Main Methods:

• Acquired 12 pairs of CDI and Diffusion Tensor Imaging (DTI) image datasets from porcine hearts.
• Applied feature-based data fusion techniques: Joint Independent Component Analysis (jICA), Canonical Correlation Analysis (CCA), and a combination (CCA+jICA).

Main Results:

• Data fusion methods enhanced and improved the classification of cardiac states derived from CDI and DTI.
• The CCA+jICA method showed a 38% increase in mean correlations compared to original images.
• Mean mutual information increased approximately three-fold for fused images from jICA and CCA+jICA.

Conclusions:

• MRI-based techniques, particularly CDI and DTI combined with data fusion, offer a more comprehensive approach to studying cardiac arrhythmias.
• These methods present viable tools for advancing research into conditions like Ventricular Fibrillation.