243Quality assurance metrics for routine clinical PET rubidium-82 myocardial blood flow quantification

Introduction: Rubidium-82 (Rb82) PET myocardial perfusion imaging (MPI) is becoming more widely used due to superior diagnostic accuracy and availability without an onsite cyclotron. For optimal myocardial blood flow (MBF) quantification with low test-retest variability, the injected Rb82 activity should be accurate, precise and delivered consistently over a short time interval. We evaluated PET MBF data obtained in a clinical setting using constant-activity-rate infusions from a new Rb82 elution system to identify useful quality assurance parameters and establish their normal limits. Methods: 4,006 patients underwent rest-stress dynamic 3D PET imaging (8,012 scans) using 10 MBq/kg, 30s constant-activity-rate ‘square-wave’ infusions of Rb82 over a 2.5-year period (680 imaging days). To obtain regional MBF estimates, a 1-tissue-compartment model was fit to the dynamic image data from 0-6 min after injection. MBF image quality was determined from the model goodness-of-fit (R²) polar-maps and K1 SNR = mean/standard deviation (SD), and evaluated vs. patient weight. Reliability of the MBF estimates was assessed using two derived quality assurance metrics to identify outlier values: i) total rest + stress K1 coefficient-of-variation (COV = SD/mean), and ii) log(rest/stress K1 COV). Results: Kinetic model R² values were excellent on average: 0.98 ± 0.2 at rest and 0.98 ± 0.3 at stress, indicating that there was high quality fitting of the model across the population sample. SNR values were excellent, with slightly higher image quality at rest vs. stress (61 ± 18 vs 47 ± 15, p < 0.001). There was no correlation of image quality with weight, indicating that uniform image quality was maintained regardless of patient size. Total rest + stress K1 COV < 10% was determined to be a useful limit to identify outliers with potential modeling or image quality issues; 14 patients (0.35%) were outside the median + 5 × IQR range using this criterion. For the log(rest/stress K1 COV), the population median ± 3 × IQR = [-1.65, 1.15] was found to be robust for identifying outliers, with only 0.27%, or 11 patients falling outside this range. QA evaluation of the outliers revealed that the majority of the dynamic images contained patient motion, suggesting that the resultant MBF (K1) values may not reliable in these cases. Conclusion: A 1-compartment kinetic model applied to dynamic Rb82 PET data obtained with 30s constant-activity-rate ‘square-wave’ infusions results in uniformly high-quality estimates of K1 and MBF. With the weight-based 10 MBq/kg dosing, image quality is consistently high over a wide range of patient weights. Total rest + stress K1 COV < 10% and log(rest/stress K1 COV) in the range of [-1.65, 1.15] are useful quality assurance metrics to identify outliers requiring investigation of potential technical issues, including patient motion. Abstract 243 Figure. Supporting Figure

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.