FACTOR-ANALYSIS FOR EXTRACTION OF BLOOD TIME-ACTIVITY CURVES IN DYNAMIC FDG-PET STUDIES

Arterial sampling in dynamic PET studies can be eliminated by using left ventricular or aortic time-activity curves (TAC) obtained from user drawn regions of interest (ROIs) after appropriate spillover correction. In this study, we evaluated the feasibility of extracting the ''pure'' arterial TAC from dynamic PET images using factor analysis of dynamic structures (FADS). Methods: Computer simulations were used to study the performance of the FADS algorithm with positivity constraints. Ten canine N-13-ammonia and two human FDG-PET dynamic studies were used to extract the blood TACs from FADS. Plasma samples and compartmental model fittings were used to validate the accuracy of the FADS-generated blood factors. Results: We found that FADS with positivity constraints was sufficient to extract the blood factor from the composite dynamic images. The ''pure'' blood-pool TACs that matched well with the arterialized well counter measurements were generated from FADS in the canine and human studies. Conclusion: FADS has the potential to accurately extract ''pure'' blood TAC from dynamic PET images, allowing reliable quantitation of biological information from PET studies without blood sampling, ROI drawing or spillover correction.