Image registration and analysis for quantitative myocardial perfusion: application to dynamic circular cardiac CT

Area of Science:

• Medical Imaging
• Cardiovascular Imaging
• Image Processing

Background:

• Large-area detector computed tomography (CT) systems facilitate dynamic cardiac perfusion studies.
• Prospective electrocardiogram (ECG)-triggered acquisitions reduce radiation dose and motion artifacts but can still suffer from cardiac cycle variations.
• These variations lead to spatial misalignments, compromising the accuracy of voxel-wise myocardial perfusion analysis.

Purpose of the Study:

• To develop and evaluate an image-based solution for spatial misalignment in dynamic cardiac CT perfusion data.
• To improve the accuracy of quantitative myocardial perfusion parameters through image registration.

Main Methods:

• Elastic 3D image registration was applied to dynamic cardiac volume sequences obtained from large-area detector CT.
• Circular cone-beam CT reconstruction was used for cardiac volumes covering myocardial tissue.
• The aligned sequences were quantitatively evaluated against unaligned data using pig datasets.

Main Results:

• Elastic 3D image registration successfully aligned cardiac volumes acquired across different cardiac cycles.
• Quantitative analysis demonstrated improved accuracy in myocardial perfusion parameters after registration.
• The method proved effective across three different pig datasets.

Conclusions:

• Elastic 3D image registration is a viable and effective method to address spatial misalignments in dynamic cardiac CT perfusion imaging.
• This technique enhances the reliability of quantitative myocardial perfusion analysis, supporting better clinical diagnosis.