Lesion detectability and quantification in PET/CT oncological studies by Monte Carlo simulations

The aim of this work was to assess lesion detectability and quantification in whole body oncological /sup 18/F-FDG studies performed by a state-of-the-art integrated Positron Emission Tomograph/computed tomography (PET/CT) system. Lesion detectability and quantification were assessed by a Monte Carlo (MC) simulation approach as a function of different physical factors (e.g., attenuation and scatter), image counting statistics, lesion size and position, lesion-to-background radioactivity concentration ratio (L/B), and reconstruction algorithms. The results of this work brought to a number of conclusions. The MC code PET-electron gamma shower (EGS) was accurate in simulating the physical response of the considered PET/CT scanner (>90%). PET-EGS and patient-derived phantoms can be used in simulating/sup 18/ F-FDG PET oncological studies. Counting statistics is a dominant factor in lesion detectability. Correction for scatter (from both inside and outside the field of view) is needed to improve lesion detectability. Iterative reconstruction and attenuation correction must be used to interpret clinical images. Re-binning algorithms are appropriate for whole-body oncological data. A MC-based method for correction of partial volume effect is feasible. For the considered PET/CT system, limits in lesion detectability were determined in situations comparable to those of real oncological studies: at a L/B=3 for lesions of 12 mm diameter and at a L/B=4 for lesions of 8 mm diameter.