Determination of extracellular/intracellular fluid ratios from magnetic resonance images: accuracy, feasibility, and...

This study determines the accuracy and feasibility of using localized spin-lattice (T1) relaxation time measurements from magnetic resonance (MR) images to follow changes in extracellular/intracellular fluid ratios in defined subvolumes of living tissue. A red blood cell suspension was used as a test system and a simple two-compartment model incorporating fast exchange was found to suffice for the conversion of T1 values to volume ratios. The technique requires the addition of gadolinium-DTPA to the model system to selectively enhance relaxation in the extracellular fluid space. No detectable amount of gadolinium-DTPA was found to enter the intracellular fluid space, and all magnetization decay plots obtained from both intracellular constituents and complete RBC suspensions consisted of a single exponential. Both of these results are compatible with assumptions underlying our physical model. The NMR-determined fluid ratio values were compared to those measured via the microhematocrit technique. Partial saturation image-mode determinations are strongly correlated to microhematocrit data (R2 = 0.945) and indicate that localized cell volume changes may be followed with a sensitivity of +/- 2.2%. These values compare favorably with those produced when nonimaging inversion-recovery techniques are used to determine the MR hematocrit (R2 = 0.962, sensitivity = +/- 1.1%). This technique, with modification, should be applicable to the comparison of ratios of extracellular/intracellular fluid volumes in structurally complex tissues where small subvolumes of homogeneous cell structure could be examined.