Objective: This study aimed to assess which region of interest results in the best intraobserver reproducibility of mebrofenin clearance rate (MCR) processing.
Methods: Mebrofenin clearance rate was calculated according to Ekman’s formalism. Processing methods included smaller left ventricular (LV) versus larger regions, nearly the whole heart (WH), which were applied to either anterior (ANT) or geometric mean (GEO) data. Each variation was processed twice by an expert operator, and Pearson linear correlation (r) was used to test for reproducibility.
Results: Studies from 50 patients (16 women) were available for processing. Mean (SD) for age, model for end-stage liver disease, and Child scores were 54 (10) years (range, 23–70 y), 13.1 (4.9) (range, 6–30), and 7.8 (2.1) (range, 5–14), respectively. The linear correlation equations (r, significance) between the 2 processing takes were y = 0.34x + 4.3 (r = 0.5442, P < 0.0001) for MCR-LV-ANT, y = 0.94x + 0.33 (r = 0.9464, P < 0.0001) for MCR-LV-GEO, y = 0.87x + 0.50 (r = 0.9760, P < 0.0001) for MCR-WH-ANT, and y = 0.92x + 0.24 (r = 0.9898, P < 0.0001) for MCR-WH-GEO. The MCR-WH-GEO method had statistically better reproducibility when compared with the other methods.
Conclusions: The highest reproducibility was found with the technique that used geometric mean data and the larger blood pool region of interest. Further work should address training approaches that foster optimal interoperator reproducibility, which could allow for wider implementation of this liver function parameter.