Kim Vang Hansen, Ole Lajord Munk & Lars Poulsen TolbodSeptember 2020

A group of physicists, engineers and technicians support the overall data flow, starting from the actual acquisition to the managing, and archiving of the reconstructed and quantitative images.

Image analysis & PET modelling

The aim of the PET kinetic modelling is to support the PET research projects with best possible analysis tools to evaluate the physiological and kinetic behaviour of various PET tracers in the body. Physiological parameters, e.g. perfusion, oxygen and glucose consumption, enzyme activity or receptor density, can be quantified by mathematical models that relate the activity concentration in tissue to that in blood supplying the tissue. A typical data set consists of the time course of the activity concentration in tissue (measured by dynamic PET), and the tracer supply to the tissue is measured in arterial blood/plasma samples that are corrected metabolites and for delay and dispersion during blood sampling. The quantification of the PET data can either be done pixel-by-pixel, producing parametric images, or regionally from tissue ROI time-activity curves. In some projects, the invasive blood sampling can be replaced by methods based on image-derived input function or reference region methods.

The typical image analysis for a brain scan consists of PET-MRI co-registration, Automatic segmentation of csf, gm, wm, wb, left, right and Atlas based co-registration and re-slicing of images. Motion correction and TAC (time activity curve) extraction can also be applied to aid the actual PET modelling.

Dedicated software is necessary to process the PET image files. Both commercial and free, even open source tools are available, such as PMOD, Minc-tools, aQuant, MATLAB, IDL, SPM, and FMRIstat. PET scanners are delivered with basic image analysis tools.


Revised: 24 September 2020