Simplified kinetic analysis of tumor 18F-FDG uptake: a dynamic approach.
Academic Article
Overview
abstract
UNLABELLED: Standardized uptake value (SUV) is often used to quantify (18)F-FDG tumor use. Although useful, SUV suffers from known quantitative inaccuracies. Simplified kinetic analysis (SKA) methods have been proposed to overcome the shortcomings of SUV. Most SKA methods rely on a single time point (SKA-S), not on tumor uptake rate. We describe a hybrid between Patlak analysis and existing SKA-S methods, using multiple time points (SKA-M) but reduced imaging time and without measurement of an input function. We compared SKA-M with a published SKA-S method and with Patlak analysis. METHODS: Twenty-seven dynamic (18)F-FDG scans were performed on 11 cancer patients. A population-based (18)F-FDG input function was generated from an independent patient population. SKA-M was calculated using this population input function and either a short, late, dynamic acquisition over the tumor (starting 25-35 min after injection and ending approximately 55 min after injection) or dynamic imaging 10 or 25 min to approximately 55 min after injection but using only every second or third time point, to permit a 2- or 3-field-of-view study. SKA-S was also calculated. Both SKA-M and SKA-S were compared with the gold standard, Patlak analysis. RESULTS: Both SKA-M (1 field of view) and SKA-S correlated well with Patlak slope (r > 0.99, P < 0.001, and r = 0.96, P < 0.001, respectively), as did multilevel SKA-M (r > 0.99 and P < 0.001 for both). Mean values of SKA-M (25-min start time) and SKA-S were statistically different from Patlak analysis (P < 0.001 and P < 0.04, respectively). One-level SKA-M differed from the Patlak influx constant by only -1.0% +/- 1.4%, whereas SKA-S differed by 15.1% +/- 3.9%. With 1-level SKA-M, only 2 of 27 studies differed from K(i) by more than 20%, whereas with SKA-S, 10 of 27 studies differed by more than 20% from K(i). CONCLUSION: Both SKA-M and SKA-S compared well with Patlak analysis. SKA-M (1 or multiple levels) had lower variability and bias than did SKA-S, compared with Patlak analysis. SKA-M may be preferred over SUV or SKA-S when a large unmetabolized (18)F-FDG fraction is expected and 1-3 fields of view are sufficient.
