RESUMEN
RATIONALE AND OBJECTIVES: To evaluate the interchangeability of perfusion parameters between two software packages for the postprocessing of dynamic contrast-enhanced (DCE) computed tomographic images of head and neck tumors. MATERIALS AND METHODS: DCE computed tomographic images of 75 patients with head and neck tumors were postprocessed using a software package based on the maximum-slope approach and Patlak analysis, as well as a software package with deconvolution-based analysis incorporating an adiabatic approximation of tissue homogeneity (ATH) model. The evaluated perfusion parameters included blood flow (F), blood volume (v), and permeability-surface area product (PS). Region-of-interest (ROI) analysis of the tumors and the metastatic lymph nodes was performed. The perfusion parameters were compared using the Wilcoxon matched-pairs test and Bland-Altman plots. RESULTS: One hundred fifty-two ROIs of tumors and nodes were outlined and analyzed. Moderate to good correlations were demonstrated between the various perfusion values (r = 0.56-0.72, P < .0001). The Wilcoxon test revealed a significant difference between the two methods (P < .001), with the F, v, and PS values obtained using the maximum-slope approach and Patlak analysis higher than those obtained using deconvolution-based analysis with the assumptions of the ATH model. The Bland-Altman plots for F and v values revealed a proportionality trend with outliers, which were strongly associated with the magnitudes of the parameters. Analysis of the PS values did not show any systematic bias. CONCLUSION: There were significant differences in the perfusion parameters obtained using the two software packages, and thus, these parameters are not directly interchangeable.