Enhancement of intra-cardiac flow-field data using adaptive Kernel filtering

Area of Science:

• Fluid Dynamics
• Medical Imaging
• Biomedical Engineering

Background:

• Vortex-dominated flow fields in cardiac chambers are vital for understanding heart function.
• Noise in flow field data, particularly from MRI, can obscure important vortex structures.
• Accurate identification of cardiac vortices is essential for disease diagnosis and prognosis.

Purpose of the Study:

• To develop a method for determining the optimal kernel size for median filtering in vortex-dominated flow fields.
• To establish a relationship between optimal kernel size and vortex size for noise reduction.
• To validate the method's effectiveness in revealing cardiac vortex patterns for clinical applications.

Main Methods:

• Generated synthetic flow fields using harmonic functions and perturbed them with Gaussian noise.
• Systematically tested the effect of median filter kernel size on noise removal.
• Applied the optimal kernel size determination method to MRI-generated cardiac flow fields.
• Correlated optimal kernel size with vortex size in both synthetic and real cardiac data.

Main Results:

• An optimal kernel size was identified for median filtering, significantly improving noise removal.
• The optimal kernel size was found to be directly related to the size of the vortices present.
• Filtering cardiac flow fields with the optimal kernel size effectively revealed underlying vortex patterns.

Conclusions:

• The developed method provides an effective approach for optimizing median filter kernel size in cardiac flow field analysis.
• Optimal kernel size is a reliable indicator related to cardiac vortex size.
• This technique serves as a valuable tool for diagnosing and prognosing cardiac diseases using vortices as biomarkers.