Table 5.

Classification of PVC methods (RB = region-based; VB = voxel-based).

Category	Method	Definition	Benefits	Limitations	Examples
RB	Recovery coefficients (RC)	They are numerical calculated in the image domain as a ratio of measured radioactivity concentration and actual concentration using spheres filled with a known radioactivity concentration	Fairly simple Practical Popular within clinical trials	More related to oncology	[369,370]
	PET raw data	Instead of using the image domain, the sinogram is used	Low computational cost	Some methods require segmentation	[371], [372], [373]
	Geometric transfer matrix (GTM) method	Average uptake is estimated considering multiple ROIs using co-registered anatomical images	Commonly used Good accuracy	Requires segmentation Bias	[374], [375], [376], [377]
VB	Image reconstruction	Spatial resolution is recovered within the image reconstruction process. It is highly recommended to use iterative reconstruction algorithms. The most relevant drawback of this is the high number of iterations needed	Increase the quality of reconstruction	Computational cost	[378], [379], [380], [381]
	Image deconvolution	Spatial resolution is recovered by using a post-reconstruction restoration technique (deconvolution) that applies the point spread function (PSF). Iterative algorithms are commonly used	Compensation of spill-over effects	Computational cost	[382]
	Multi-resolution approach	Spatial resolution is recovered by using information from spatially coregistered high-resolution anatomical images, which allows transfer high spatial frequencies from anatomical to functional images	Adjust of radioactivity concentration Used in brain imaging	In poorly correlated areas the amount of artifacts increases	[383]
	Anatomical images	First, a segmentation of a coregistered anatomical image is done. Then, PVE correction is applied to the set of voxels associated with each region/class (tissue)	Highly used in brain imaging	Need segmentation	Two classes: [364,365] Three classes: [366] Four classes: [367]