Ultrafast magnetic-resonance-imaging velocimetry of liquid-liquid systems: overcoming chemical-shift artifacts using compressed sensing.

We present simultaneous measurement of dispersed and continuous phase flow fields for liquid-liquid systems obtained using ultrafast magnetic resonance imaging. Chemical-shift artifacts, which are otherwise highly problematic for this type of measurement, are overcome using a compressed sensing based image reconstruction algorithm that accounts for off-resonant signal components. This scheme is combined with high-temporal-resolution spiral imaging (188 frames per second), which is noted for its robustness to flow. It is demonstrated that both quantitative signal intensity and phase preconditioning are preserved throughout the image reconstruction algorithm. Measurements are acquired of oil droplets of varying viscosity rising through stagnant water. From these data it is apparent that the internal droplet flow fields are heavily influenced by the droplet shape oscillations, and that the accurate modeling of droplet shape is of critical importance in the modeling of droplet-side hydrodynamics. The application of the technique to three-component systems is also demonstrated, as is the measurement of local concentration maps of a mutually soluble species (acetone in polydimethylsiloxane-water).