4D-CT-based motion correction of PET images using 3D iterative deconvolution

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Radiology

Background:

• Positron Emission Tomography (PET) imaging involves multi-minute acquisitions, averaging respiratory motion.
• Respiratory motion artifacts in PET lead to inaccurate lesion size and intensity, compromising quantitative assessment.
• Accurate lesion characterization is crucial for effective diagnosis and treatment planning in oncology.

Purpose of the Study:

• To develop and validate a motion-correction algorithm for PET imaging using 4D-CT data.
• To improve the quantitative accuracy of lesion volume and activity uptake in PET scans affected by respiratory motion.
• To assess the efficacy of the algorithm in both phantom studies and patient data.

Main Methods:

• Developed a 3D iterative Richardson-Lucy-Deconvolution algorithm incorporating motion information from 4D-CT.
• Estimated optimal algorithm parameters (iterations, start image) using a motion phantom with varying sphere sizes.
• Applied the motion-correction method to three patient datasets with lung lesions, comparing corrected and uncorrected images.

Main Results:

• Phantom studies demonstrated optimal motion correction performance with 6 or more deconvolution steps.
• Lesion volume reduction in phantom studies ranged from 23% to 49% across different sphere sizes.
• Patient data showed significant tumor volume reduction (up to 33.3%) and increased lesion uptake (max: 62.1%, mean: 19.8%) after correction.

Conclusions:

• The developed motion-correction algorithm effectively reduces lesion volume in PET imaging.
• The method shows promising improvements in quantitative accuracy, increasing lesion activity uptake.
• This technique offers a valuable tool for more precise assessment of lung lesions in PET scans.