Abstract 1974: Correcting Curvilinear Myocardial Signal Response to Gadolinium Improves Accuracy of Absolute Myocardial Blood Flow Quantification by MRI Perfusion Imaging
Background: Absolute myocardial blood flow (in ml/min/g) can be measured from MRI first-pass perfusion studies by deconvolution of signal intensity-time curves from the LV blood pool (LV) and myocardium (Myo). This technique assumes signal intensity (SI) is proportional to gadolinium concentration [Gd] both in Myo and LV. The dual bolus technique can circumvent the known nonlinearity of signal in LV, but Myo nonlinearity must also be addressed.
Purpose: We sought to:
measure the relationship between [Gd] and SI in Myo and LV;
use this relationship to convert SI-time curves to [Gd]-time curves; and
compare flow values derived from SI-time curves and [Gd]-time curves to microsphere-derived flow.
Methods and Results: Studies were performed in chronically instrumented dogs on a 1.5 T clinical MRI scanner.
The SI vs [Gd] relationship in Myo and LV were derived by measuring SI using a first-pass sequence and calculating regional [Gd] from T1 measured by a modified Look-Locker sequence during continuous Gd infusion. Myo was curvilinear, plateauing over the signal range of a 0.05 mmol/kg Gd bolus while LV was linear over the range of a 0.005 mmol/kg minibolus.
First-pass studies were performed with dual bolus (0.005/0.05 mmol/kg) at rest and during vasodilation with variable degrees of left circumflex stenosis. SI-time curves were converted to [Gd]-time curves using the relationship derived above.
Regional myocardial blood flow was calculated by deconvolution of SI-time curves and [Gd]-time curves. In comparison to reference microspheres, SI-time curves underestimated flow (y = 0.48x − 0.1, r2 = .70) while [Gd]-time curves were more accurate (y = 0.94x − 0.3, r2 = .79).
Conclusions: The relationship between SI and [Gd] in the myocardium is nonlinear and plateaus over the Gd dose range utilized in clinical MR perfusion studies. Using a dual bolus strategy to address LV nonlinearity while ignoring myocardial nonlinearity results in significant underestimation of myocardial blood flow by deconvolution. By measuring the SI vs. [Gd] relationship, SI-time curves can be converted to [Gd] time curves. Deconvolution of [Gd]-time curves significantly improves the accuracy of absolute myocardial blood flow measurement by MRI first-pass perfusion imaging.