Purpose: Dynamic susceptibility contrast MRI (DSC) and dynamic contrast-enhanced MRI (DCE) are useful tools in the diagnosis and follow-up of brain gliomas; nevertheless, both techniques leave the open issue of data reproducibility. We evaluated the reproducibility of data obtained using two different commercial software for perfusion maps calculation and analysis, as one of the potential sources of variability can be the software itself. Methods: DSC and DCE analyses from 20 patients with gliomas were tested for both the intrasoftware (as intraobserver and interobserver reproducibility) and the intersoftware reproducibility, as well as the impact of different postprocessing choices [vascular input function (VIF) selection and deconvolution algorithms] on the quantification of perfusion biomarkers plasma volume (Vp), volume transfer constant (Ktrans) and rCBV. Data reproducibility was evaluated with the intraclass correlation coefficient (ICC) and Bland–Altman analysis. Results: For all the biomarkers, the intra- and interobserver reproducibility resulted in almost perfect agreement in each software, whereas for the intersoftware reproducibility the value ranged from 0.311 to 0.577, suggesting fair to moderate agreement; Bland–Altman analysis showed high dispersion of data, thus confirming these findings. Comparisons of different VIF estimation methods for DCE biomarkers resulted in ICC of 0.636 for Ktrans and 0.662 for Vp; comparison of two deconvolution algorithms in DSC resulted in an ICC of 0.999. Conclusions: The use of single software ensures very good intraobserver and interobservers reproducibility. Caution should be taken when comparing data obtained using different software or different postprocessing within the same software, as reproducibility is not guaranteed anymore. © 2017, Italian Society of Medical Radiology.
|Number of pages||9|
|Publication status||Published - 2017|