Magnetic resonance imaging multicompartment perfusion model: theory and experimental verification

A multicompartment model of MRI signal intensity that is a function of perfusion is developed based upon the assumption that biological tissue can be represented by blood, tissue and immobile water compartments and that excited endogenous protons can be used as a tracer. The principle is analogues to tracer kinetic techniques used in many fields of biological science. First, the longitudinal magnetization for a two-compartment model, representing blood and tissue, is derived from the modified Bloch equations as a function of the following physiological parameters: blood flow velocity, tissue to blood volume fraction, diffusion, and rate of exchange between the blood and extravascular tissue compartments. Simulations of slice profiles excited by a repetitive sequence of 90° slice-selective pulses show that the signal intensity in the compartments are modulated by these physiological parameters. Second, the longitudinal and transverse magnetization for both a two-compartment model and a three-compartment model are derived and studied using chromatography column phantoms containing Sephadex gels, which were used to simulate tissue perfusion and the exchange of protons between extravascular and intravascular tissue compartments. Computer simulations were compared in experiments that used two chromatography columns. Slice-selective spin-echo experiments were performed. The results of the experiments agreed with computer simulations, which showed that the MRI signal intensity in the perfused columns is a function of the rate of exchange between extrabead and intrabead compartments. The exchange process modifies the transit time of protons passing through an excited region. Simulations and experiments also showed that both two-compartment and three-compartment models could be used to fit experimental data. Finally, an experiment was performed on a human brain using arterially tagged endogenous protons as a tracer combined with magnetization transfer techniques to eliminate the immobile water compartment. Our simulations and experimental results show that the accuracy of kinetic parameter estimates will rely on the signal contrast that depends upon the flow velocity of the labeled spins. The arterial spin labeling technique has significant potential to be used for quantitatively measuring tissue perfusion in vivo using clinical MRI.