Partial volume correction of brain PET studies using iterative deconvolution in combination with HYPR denoising

Area of Science:

• Nuclear Medicine
• Medical Imaging
• Image Processing

Background:

• Accurate quantification in PET imaging is crucial but limited by spatial resolution, leading to partial volume effects (PVE).
• Iterative deconvolution methods (IDM) are used to correct PVE and improve spatial resolution without requiring structural information.
• However, IDM can increase image noise, reducing signal-to-noise ratio (SNR).

Purpose of the Study:

• To implement IDM combined with HighlY constrained back-PRojection (HYPR) denoising.
• To mitigate the poor SNR associated with conventional IDM.
• To evaluate the impact of this combined approach on quantitative accuracy in dynamic PET studies.

Main Methods:

• An anthropomorphic brain phantom was used to optimize IDM parameters and HYPR implementation (before/after IDM).
• Dynamic PET scans ([18F]FDG, [11C]flumazenil, [11C]PIB) were acquired on a Philips Ingenuity TF PET/CT scanner.
• The optimized HYPR-IDM-HYPR approach was assessed for its effect on distribution volumes (VT) in various brain regions.

Main Results:

• The HYPR-IDM-HYPR method demonstrated improved quantitative accuracy for measured activity concentrations.
• [11C]flumazenil scans showed a 5-20% increase in regional VT.
• [11C]PIB scans showed a 0-10% increase or decrease in regional VT, depending on the region and subject type.

Conclusions:

• IDM effectively enhances the quantitative accuracy of PET measurements.
• The combination of IDM with HYPR (HYPR-IDM-HYPR) successfully corrects for PVE.
• This combined approach corrects for PVE without introducing additional noise, preserving SNR.