Limitations in biexponential fitting of NMR inversion-recovery curves.

NMR relaxation agents have long been employed as contrast agents in MRI. In many cases, the contrast agent is confined to either (i) the vascular and/or extracellular compartment (EC), as is the case with gadolinium(III)-based agents, or (ii) the intracellular compartment (IC), as is the case with manganese(II) ions. The compartmentalization of contrast agents often results in tissue-water 1 H relaxation profiles that are well modeled as biexponential. It has long been recognized that water exchange between compartments modifies the biexponential relaxation parameters (amplitudes and rate constants) from those that would be found in the absence of exchange. Nevertheless, interpretation in terms of an "apparent" two-compartment biophysical model, apparent EC vs. apparent IC, can provide insight into tissue structure and function, and changes therein, in the face of physiologic challenge. The accuracy of modeling biexponential data is highly dependent upon the amplitudes, rate constants, and signal-to-noise characterizing the data. Herein, simulated (in silico) inversion-recovery relaxation data are modeled by standard, nonlinear-least-squares analysis and the error in parameter values assessed for a range of amplitudes and rate constants characteristic of in vivo systems following administration of contrast agent. The findings provide guidance for laboratories seeking to exploit contrast-agent-driven, biexponential relaxation to differentiate MRI-based compartmental properties, including the apparent diffusion coefficient.