Investigation of partial volume correction methods for brain FDG-PET studies [MRI application]

The usage of PET in quantitative FDG studies of aging and dementia has been limited by partial volume effects. A general method for correction of partial volume effects (PVE) in PET involves the following common procedures: (1) segmentation of MRI brain images into gray matter (GM), white matter (WM), cerebral spinal fluid (CSF), and muscle (MS) components, (2) MRI-PET registration, (3) generation of simulated PET images. Afterward, two different approaches can be taken. Approach A derives first a pixel-by-pixel correction map as the ratio of the measured image to the simulated image (with realistic FWHM). The correction map was applied to the MRI segmentation image. ROIs can then be applied to give results free of partial volume effects. Approach B uses the ROI values of the simulated "pure" image (with negligible FWHM), and those of the simulated and the measured PET images to correct for the PVE effect. By varying the ratio of radiotracer concentrations for different tissue components, the in-plane FWHMs of a 3-dimensional point spread function, and the ROI size, the authors evaluated the performance of these two approaches in terms of their accuracy and sensitivity to different simulation configurations. The results showed that both approaches are more robust than the approach developed by Muller-Gartner, et al. (1992), and approach B is more accurate and more robust than approach A. In conclusion, the authors recommend that approach B should be used on FDG PET images to correct for partial volume effects and to determine whether an apparent change in GM radiotracer concentration is truly due to pathophysiological process.