Utilization of CT images for the quantification of FDG uptake

The aim of this study was to evaluate an automatic method based on a computed tomography (CT) derived region of interest (ROI) to quantify the mean standardized uptake value (SUVm) of 18F-fluoro-deoxy-glucose (FDG) in pulmonary lesions detected by positron emission tomography (PET). A total of 164 pairs of slices were selected in a series of PET/CT studies performed in 26 patients presenting lung tumours of various forms and complexities. On each matched CT slice, a ROI was obtained by growth-region segmentation starting from a pixel contained in the tumour. The obtained ROI was then applied to the PET image to calculate SUVm. Results were compared with the conventional manual method using a geometric ROI positioned directly on the PET lesion. The automatic delineation of the tumour from the CT image was successful in 136 sections (83%). The SUVm calculated by the manual and automatic method were respectively (mean±standard deviation) 5.05±2.39 and 6.70±3.18 (p<0.05). The ROI size (in number of pixels) was respectively 28±23 and 21±17 (p<0.05). The variability of the automatic method was 0% versus 20% for the manual method. SUV of FDG in PET/CT can be calculated with an excellent reproducibility by using the CT-derived limits of the lesion.