Lesion quantification in oncological positron emission tomography: a maximum likelihood partial volume correction...

Area of Science:

• Nuclear Medicine
• Medical Imaging Analysis
• Quantitative Oncology

Background:

• Accurate quantification of oncological lesions in 18F-FDG PET is crucial for treatment assessment.
• Partial volume effects (PVE) in PET imaging can lead to underestimation of lesion activity and volume.
• Existing correction methods may lack accuracy or computational efficiency.

Purpose of the Study:

• To develop and evaluate a novel maximum likelihood (ML) partial volume effect correction (PVEC) strategy for 18F-FDG PET.
• To improve the accuracy of lesion uptake and volume quantification in oncological imaging.
• To assess the computational efficiency of the proposed PVEC method.

Main Methods:

• A ML-based PVEC strategy using iterative reconstruction and segmentation was developed.
• The algorithm employs volumetric regional basis functions updated iteratively for activity and volume.
• Segmentation uses a k-means algorithm defining central, partial volume, and spill-out regions within the VOI.
• An attenuation-weighted ordered subset expectation maximization (AWOSEM) algorithm with point spread function (PSF) recovery was utilized.

Main Results:

• The PVEC strategy demonstrated improved accuracy in estimating lesion volume and activity compared to conventional methods.
• Phantom studies showed enhanced quantification with activity contrasts of 7.5 and 4.
• The method achieved improved results with low computational costs.

Conclusions:

• The proposed ML-based PVEC strategy offers a promising approach for accurate oncological lesion quantification in 18F-FDG PET.
• The integration of resolution recovery and advanced segmentation improves local approximation to ML convergence.
• This method provides a computationally efficient solution for PVE correction in PET imaging.