Partial Volume Correction in PET

Ideally, after corrections for the physical artifacts (e.g. scatter and attenuation) and calibration, the reconstructed PET image should represent highly accurate radiopharmaceutical distribution in absolute units of radioactivity concentration throughout the field of view of the PET scanner. However, this only holds for organs or structures with dimensions greater than twice the spatial resolution of the scanner, which is characterized by the full width at half maximum height of an image of a line source. When the object or structure being imaged is smaller than this, the apparent activity concentration is diluted. The degree of dilution in activity concentration varies with the size of the structure being imaged and the radioactivity concentration of the imaged structure comparing to its surrounding structures [10]. This phenomenon is known as partial volume effect (PVE), which is solely caused by the limited spatial resolution of the PET scanner.

A number of approaches have been proposed to correct or minimize the PVE, including resolution recovery before or during image reconstruction, and incorporation of side information provided from anatomical imaging modalities such as CT and MRI. One of the popular approaches that incorporates MRI segmentation for partial volume correction is the method proposed by Muller-Gartner et al. [54] but the method is only applicable to brain imaging. PET images are first spatially co-registered with MR images obtained from the same subject. The MR images are then segmented into gray matter, white matter, and CSF regions, represented in three separate images. These images are then convolved spatially with a smoothing kernel which is derived from the point spread function of the PET scanner. The convolved white matter, image is then normalized to the counts in a white matter ROI drawn on the PET image so that spillover white matter activity into gray matter regions can be removed. Finally, the resultant image is divided by the smoothed gray matter image so that signals in small structures, which were smoothed severely, can be enhanced.

0 0

Post a comment