…but we can still calculate true concentrations if we have enough information!
A-posteriori PVE correction is achievable when: (1) the distribution map of the various structures within the functional images can be accurately defined (i.e. domains Di are correctly delineated) and (2) the resolution of the scanner h(x) is known. High-resolution anatomical images, such as those provided by Magnetic Resonance Imaging (MRI), contain the morphological information needed to delineate the distribution map of structures. On the other hand, the functional image resolution can either be calculated knowing the intrinsic resolution of the scanner and the reconstruction filter, or measured by scanning simple geometric phantoms.
A-posteriori PVE corrections can then be performed according to the calculated output of the scanner when scanning a known object. This can be imagined as a map, corresponding to a PET image, where each voxel has a value equal to the relative volume of the object that contributes to that PET voxel. We will refer to this procedure hereinafter as "virtual PET".
 |
Example
of a WM Virtual PET, obtained simulating the PET image formation
process starting from a theoric tracer distribution map (the “base”
object, in this case the bitmap of the WM obtained from a segmented MR
image). This process can be can be performed with different degrees of accuracy of the simulation ranging from a simple 2D-smoothing or by a more or less complete simulation of the PET process |
MQ
| Home |