RT Journal Article
SR Electronic
T1 Effects of noise, calibration, reconstruction and analysis methods on correctly determining change in PET SUVs
JF Journal of Nuclear Medicine
JO J Nucl Med
FD Society of Nuclear Medicine
SP 1772
OP 1772
VO 52
IS supplement 1
A1 Kinahan, Paul
A1 Harrison, Robert
A1 Elston, Brian
A1 Doot, Robert
A1 Lewellen, Tom
A1 Mankoff, David
YR 2011
UL http://jnm.snmjournals.org/content/52/supplement_1/1772.abstract
AB 1772 Objectives PET/CT imaging is increasingly being used for quantitative assessment of individual response to therapy and for clinical trials of novel cancer therapies. It is thus important to understand the ability of quantitative PET/CT to correctly determine if measured changes in PET standardized uptake values (SUVs) reflect changes in the true PET tracer uptake. We use simulation studies to evaluate the probability of false-positive and false-negative errors as a function of measurement noise, calibration variability and variations in image reconstruction and analysis methods. Methods We used the ASIM simulation program to generate multiple realizations of breast PET acquisitions with accurate physics models with increased and decreased tumor uptake of FDG. Background SUV was 0.5. Parameters varied included true tumor uptake and change, scan duration, longitudinal scanner calibration variability, reconstruction method (FBP and OSEM), and analysis method (SUVmean and SUVmax). ROC analysis was used to estimate the overall area under the ROC curve (AUC). Results Multiple scenarios were evaluated. One example is a 3 cm tumor with true SUV changing from 4 to 3.6 between PET scans, reconstructed using OSEM and SUVmax, with a typical scan duration and no variations in scanner calibration or patient test-retest variability. In this case the AUC was 0.99. With added scanner variability of 3% and patient test-retest variability of 10%, AUC decreased to 0.70. In general the PERCIST classification method was less error-prone than the EORTC method. Results were much less sensitive to changes in scan duration and reconstruction method than compared to changes in scanner calibration or patient test-retest variability. Conclusions For accurate measurement of response to therapy, changes in image reconstruction method or scan duration were less important than increased variability in either scanner calibration or patient test-retest values. Research Support Supported by NIH grants R01-CA115870, R01-CA126593, R01-CA42593 and U01-CA14813