Improvement in breast lesion characterization with dynamic contrast-enhanced MRI using pharmacokinetic modeling and bookend T1 measurements

Dynamic contrast-enhanced breast MR imaging was performed on 14 patients (five cancerous lesions, nine benign) with slice-selective spoiled gradient-recalled echo (2D SPGR) imaging. Adiabatic saturation recovery T-1 measurements were performed before (T-1pre) and after (T-1post) 2D SPGR imaging. These two "book-end" T-1 measurements were used to calibrate the equations which were employed to convert the time course of the 2D SPGR signal strength to T-1-vs.-time, which in turn was used to compute the gadolinium concentration-vs.-time ([C](t)) in the lesion. The extraction-flow product (EF) was computed for each lesion by pharmacokinetic modeling of [C](t). For this study, EF provided a sensitivity and specificity for cancer of 100% and 78%, respectively. When only T-1pre was used to estimate [C](t) (which assumes a priori knowledge of the shape and amplitude of the slice profile), the sensitivity and specificity fell to 80% and 56%, respectively. This is presumably due to unexpected variations in the shape and/or amplitude of the slice profile, which could be caused by factors such as patient-to-patient variations in breast geometry or inconsistently set transmit gains. Therefore, both T-1pre and T-1post measurements are necessary for optimum sensitivity and specificity using pharmacokinetic analysis. (C) 2004 Wiley-Liss, Inc.