Retrospective correction of surface coil MR images using an automatic segmentation and modeling approach.

The use of surface coils in magnetic resonance imaging offers significant improvements in the signal-to-noise ratio over volume coils for many applications. However, the inhomogeneous reception profile of surface coils hampers their usefulness by introducing significant nonuniformities or intensity variations which can vary by greater than six-fold across the sample. In this study, we evaluated an automatic technique for retrospective correction of intensity variations observed in a high-resolution surface coil MR image of the rat brain obtained using an adiabatic magnetic resonance imaging sequence. The images are shown to have a coefficient of variation less than 12% following application of this correction algorithm. This image intensity correction technique can be applied retrospectively to all data sets and corrects both sample/patient dependent effects (e.g. attenuation of overlying tissue) or sample independent effects (e.g. coil geometry or position). This approach should also prove valuable in improving regions of interest analysis, volume histograms and thresholding techniques.