Attenuation-emission alignment in cardiac PET/CT based on consistency conditions

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Cardiovascular Imaging

Background:

• Cardiac PET/CT imaging is prone to misalignment between transmission and emission data due to patient motion.
• This misalignment causes inaccurate attenuation correction, leading to errors in perfusion quantification and mapping.

Purpose of the Study:

• To develop and validate an automated method for aligning transmission and emission data in cardiac PET/CT.
• To improve the accuracy of attenuation correction and subsequent perfusion analysis.

Main Methods:

• Iterative transformation of CT-based attenuation maps using Radon consistency conditions.
• Evaluation using simulated data across various noise levels and patient data from cardiac ammonia PET/CT exams.
• Quantitative assessment of alignment accuracy and perfusion accuracy on simulated and real patient data.

Main Results:

• The automated alignment procedure significantly reduced alignment errors in simulations (RMSE < 5 mm, axial translation < 1 mm).
• Patient studies showed >50% reduction in translation error and resolution of perfusion artifacts.
• Reviewers overwhelmingly preferred the automatically aligned images over initially misaligned ones.

Conclusions:

• The proposed automated alignment procedure effectively reduces attenuation correction artifacts in cardiac PET/CT.
• It serves as a valuable supplement to manual realignment techniques, enhancing diagnostic accuracy.