Target field design for MAMBA step fields

Multiple Acquisition with Micro-B0 Array (MAMBA) is a class of parallel imaging techniques. Its distinguishing feature is the use of auxiliary coil arrays that introduce a stepped field within the B0 field. The steps form regions of uniform field, which then can be used to encode in one or more dimensions. The steps can also be used in conjunction with sensitivity encoding (SENSE) techniques to achieve higher acceleration factors. Step fields necessarily contain high-order harmonics and require high currents to implement them. To reduce current demands, we have combined the pseudoinverse target field method with subset selection and Tikhonov regularization to derive coil array designs for some MAMBA configurations. We conclude that MAMBA will be most effective in long and narrow or thin and flat configurations. In these situations, our algorithm enables sparse designs with low current demands to be obtained. © 2004 Wiley Periodicals, Inc. Concepts Magn Reson Part B (Magn Reson Engineering) 20B: 1–8, 2004.