Effect of Q.Clear reconstruction on [13N]-NH3 perfusion imaging in suspected cardiac sarcoidosis

Abstract

Background Detection of cardiac involvement in patients with suspected extra-pulmonary sarcoidosis is of paramount importance in their clinical management. We evaluate the use of a new commercially available Bayesian penalized likelihood (BPL) reconstruction algorithm (Q.Clear – GE Healthcare) on dynamic [13N]-NH3 quantified perfusion data, which has to date, not yet been investigated. Methods We retrospectively analysed data from 10 patients who had undergone rest perfusion imaging with dynamic [13N]-NH3 in combination with [18F]-FDG for viability imaging. We reconstructed [13N]-NH3 datasets with a range of BPL penalization coefficients from 1-500 and calculated K1 perfusion values for all [13N]-NH3 data using commercially available software utilizing a 2-compartment kinetic model. All segmentations and calculations were automatically generated by the software. Results Reconstruction of the [13N]-NH3 dynamic data using the BPL or standard FBP showed no quantitative difference on the calculation of rest MBF in terms of territorial or global perfusion ( p = 0.98). BPL images generated less noisy time activity curves and images appeared smoother with higher penalization factor. Conclusion Quantitation of perfusion via kinetic modelling of cardiac rest perfusion by [13N]-NH3 is unaffected by the use of a BPL reconstruction technique, yet subjectively the images present with less noise and appear smoother with BPL. Although noise is not factored into the commercial software, we propose that perfusion value should be quoted with an associated error.