Assessment of the impact of post-processing approaches on absolute myocardial blood flow quantification using...

Quantitative myocardial blood flow (MBF) assessment using cardiac-dedicated cadmium-zinc-telluride (CZT) single photon emission computed tomography (SPECT) has emerged as a promising noninvasive approach for evaluating both macrovascular and microvascular coronary function. However, the reliability and comparability of MBF measurements are substantially influenced by variations in post-processing methodologies, including the application of attenuation correction (AC) and scatter correction (SC) techniques. Despite the increasing clinical adoption of CZT-SPECT for absolute perfusion quantification, systematic evaluations comparing different post-processing strategies remain limited. This study aimed to evaluate the impact of various post-processing approaches on absolute quantitative MBF measurements obtained through CZT-SPECT, with particular focus on AC and SC methodologies. Seventy patients who underwent dynamic CZT-SPECT myocardial perfusion imaging were retrospectively included, comprising 35 Tc-99m-methoxyisobutylisonitrile patients and 35 Tc-99m-tetrofosmin patients. Each raw imaging dataset was independently reprocessed using Corridor 4DM under 3 correction strategies - no correction, attenuation correction, and attenuation + scatter correction (ACSC) - as well as using MyoFlowQ. Repeated-measures analysis of variance or Friedman tests with Holm-adjusted pairwise comparisons were used to compare the Corridor 4DM processing strategies, and agreement between Corridor 4DM with ACSC and MyoFlowQ was assessed using Bland-Altman analysis. Rest and stress MBF differed significantly across the Corridor 4DM processing strategies (both P < .001). Myocardial flow reserve (MFR) differences were tracer-dependent: no overall difference was observed in global left ventricular values in the Tc-99m-methoxyisobutylisonitrile cohort (P = .121), whereas a significant difference was observed in the Tc-99m-tetrofosmin cohort (P = .002). Compared with Corridor 4DM using ACSC, MyoFlowQ yielded higher stress MBF and MFR values (bias, 1.19-1.31 mL·g-1·min-1; 95% limits of agreement, -0.30 to 2.68), with moderate inter-software correlation (r = 0.44-0.59, P < .01). Post-processing methodology significantly influenced absolute MBF quantification in CZT-SPECT imaging. Within Corridor 4DM, correction strategies affected MBF estimates, and MyoFlowQ yielded higher stress MBF and MFR values than Corridor 4DM using ACSC, indicating that these post-processing approaches are not directly interchangeable.