The use of adaptive algorithms for obtaining optimal electrical shimming in magnetic resonance imaging (MRI)

A method of determining the DC coil current values in order to shim electrically the static magnetic fields used in magnetic resonance imaging (MRI) using the modified steepest descent adaptive algorithm is described. Using a 32-cm-diameter by 40-cm-long water phantom as the test volume, the algorithm achieved field homogeneities of 0.2 parts per million (p.p.m.) peak-to-peak within a 20-cm-diameter spherical imaging volume, and 1.3 p.p.m. peak-to-peak within the entire phantom. The algorithm achieved an inhomogeneity variance of 0.18 p.p.m./sup 2/. The shim system was successfully modeled as a sum of adaptive linear combiners. The model, which contains 13 parameters that can be varied, 12 shim coil currents, and the receiver mixer frequency, has been used to predict key adaptive algorithm parameters. Experimental verification of these parameters lends support to the accuracy of the model.<>