Quantifying the normal range of myocardial blood flow with 13N-ammonia PET and tracer kinetic modeling

To aid in diagnosis of coronary artery disease (CAD), automated interpretation programs compare patient scans with a database defining the lower limit of normal flow. While absolute myocardial blood flow can be quantified using N-13-ammonia PET and various tracer kinetic models, normal databases have only been reported using compartmental modeling. This study investigated whether simplified kinetic models could reduce the measured range of normal flow and improve identification of regional flow abnormalities. 14 subjects with <5% likelihood of CAD underwent rest and dipyridamole stress N-13-ammonia dynamic PET imaging. Flow was quantified using a standard 2-compartment model. Patlak analysis was used to estimate the initial uptake rate (K). A simple first-pass model was used to estimate the net retention rate (R). Standardized flow estimates were derived from K and R using model-specific extraction functions. Uniformity of the 2-compartment model polar-maps was improved using k₂/K₁ constraints derived from subject-specific and population-average values. LV-median rest, stress and stress/rest values were similar for all flow models. Database polar-map values of regional flow were more variable than the LV-median results due to added regional noise. The net retention rate had the smallest variability, indicating that this model may be preferable for detection and localization of flow abnormalities, whereas the compartmental models are more precise for absolute flow quantification.