The feasibility of 3-dimensional acquisition mode for semiquantitative anal
ysis in thoracic PET studies was compared to the conventional 2-dimensional
mode. Several practical considerations were analyzed to propose an optimiz
ed scanning protocol for clinical use. Methods: Twenty-one patients with fo
cal thoracic abnormalities were evaluated with FDG PET. The acquisition con
sisted of 3 consecutive static scans for a single bed position: S-dimension
al (10 min), 2-dimensional (15 min), and 3-dimensional (5 min). On the basi
s of the average and maximum activity values per region of interest, standa
rdized uptake value (SUV) normalized for total body weight (TBW), lean body
mass (LBM), body surface area (BSA), and blood glucose level (PGL) were ev
aluated. The effect of the delay between tracer injection and PET scanning
on the SUV, as well as on the relative error of the activity distribution,
was studied from 40-134 min after tracer injection. Results: A strong posit
ive correlation was observed among SUVs from 2-dimensional and both 3-dimen
sional acquisitions. The mean SUV percentage differences between both acqui
sition modes were about 17%, differences that were not statistically signif
icant when time postinjection was addressed in the analysis of covariance.
SUVs provided the greatest variability and differences among studies on exp
erimental periods up to 70 min postinjection. Indeed, the variability of 20
% observed on the SUVs from 2 PET scans 13 min apart was reduced to 9% when
the acquisitions started at least 70 min after tracer injection. In additi
on, a two-fold reduction in the relative error of the activity distribution
was observed over this period of time. The reproducibility coefficient was
increased from 0.87 to 0.95 before and after 70 min postinjection, respect
ively. No correlation was found between different normalization procedures
of SUV and LBM, BSA, TEW, or height, whereas a weak correlation was found b
etween SUV and PGL, Conclusion: F-18-FDG 3-dimensional PET is a realistic a
lternative to the gold standard 2-dimensional for clinical nonkinetic studi
es. A short, 5-min 3-dimensional acquisition at 70 min postinjection is pro
posed as the best protocol for the clinical evaluation of thoracic patholog
ies.