Model-free parameters from dynamic contrast-enhanced-MRI: Sensitivity to EES volume fraction and bolus timing

• Published in: Journal of magnetic resonance imaging 2006-09, Vol.24 (3), p.586-594
• Main Authors: Jesberger, John A., Rafie, Niusha, Duerk, Jeffrey L., Sunshine, Jeffrey L., Mendez, Matthew, Remick, Scot C., Lewin, Jonathan S.
• Format: Article
• Language: English
• Subjects: angiogenesis; Contrast Media - administration & dosage; Contrast Media - pharmacology; diffusible tracers; dynamic contrast-enhanced MRI; Extracellular Space - metabolism; Gadolinium DTPA - pharmacology; Humans; Magnetic Resonance Imaging - methods; methodology; Models, Statistical; Neovascularization, Pathologic; perfusion imaging; Sensitivity and Specificity; Time Factors; Tissue Distribution
• ISSN: 1053-1807; 1522-2586
• DOI: 10.1002/jmri.20670

Purpose To quantify the unknown relative sensitivities of semiquantitative measures from dynamic contrast‐enhanced (DCE) MRI to variations in the volume fraction Ve of the extravascular extracellular space (EES), and the duration of the contrast injection. Materials and Methods Tissue‐uptake curves were simulated across various values of F, PS, Ve, and bolus timings, with and without additive noise and at different image reacquisition rates. From each, the peak of the first derivative (Gpeak), the total uptake after the rapid first phase (CE), and the IAUC were calculated and plotted against F for each experimental condition. Relationships between each measure and the corresponding quantitative measure Ktrans were also examined, particularly for linearity. Results The highest sensitivity to flow was achieved for shorter bolus timings for Gpeak, CE, and IAUC. Gpeak and IAUC were most linearly related to Ktrans. The sensitivity to Ve was lowest for Gpeak, followed by IAUC and CE. Long sampling intervals resulted in severe underestimation of Gpeak, while IAUC was unaffected provided that the limits of integration were properly applied. Gpeak could not be properly calculated in the presence of noise without a prior smoothing of the acquired curves, while IAUC was again unaffected by noise. Conclusion Gpeak and IAUC are both useful model‐free analogs of blood flow (i.e., Ktrans) for pre‐ and posttreatment comparisons. Gpeak may be the better choice in cases where larger changes in Ve are likely, but only if sufficient noise reduction and fast image sampling are applied. If Ve is expected to remain stable, IAUC is superior to Gpeak by virtue of its stability in the face of noise and more reliable estimation over a wider range of sampling rates. J. Magn. Reson. Imaging 2006. © 2006 Wiley‐Liss, Inc.