Determination of Flow Velocities from Magnetic Resonance Multiple Spin-Echo Images

The purpose of this study was to evaluate a method for the quantification of through-plane flow velocities by magnetic resonance imaging (MRI) from the flow characteristics of conventional multiple spin-echo (MSE) signals. Simulated inflow-outflow-dependent signals, as well as images of a flow phantom were generated and the logarithm of the flow-dependent signal value was plotted against echo time. The normalized slope of the resulting curve was calculated using a least-square fit to simulated and experimental data and was corrected for T2 relaxation effects by subtraction of a slope obtained at zero flow. After this correction, and with certain restrictions regarding the flow velocity (v), maximum number of echoes in the slope calculation and slice thickness (L), the normalized slope of the MSE signal becomes equal to the quotient v/L, and from this relation the flow velocity can be determined. The validity of the proposed method was examined for different flow velocities and for two opposite flow directions. The influence of the size of the region of interest and the number of spin echoes used in the calculation of the slope on the accuracy of the velocity determination was also studied. The sensitivity of the method to flow-induced phase changes was examined in the phantom by comparing the results obtained with different strengths of the slice-selective gradient as well as by comparing results from even-echo data with those from odd-echo data. When applied to simulated signal data, the method was found to be strictly valid only for a small velocity range, while for the flow phantom, the calculated velocities corresponded to measured velocities for values up to and over 100 mm/s. In the phantom experiment, the method was found to be insensitive to effects induced by combined changes of the slice thickness and the slice-selective gradient as well as to so-called even-echo rephasing effects. It is concluded that the examined method promises to be a rapid and easily interpretable alternative to other methods, e.g. magnetic resonance velocity-phase encoding, for the determination of flow velocities in vivo.