Bias Field Correction of Breast MR Images

We present a method to automatically estimate and remove the bias field of MR images where there is a single dominant tissue class. Assuming that a multi-class image is corrupted by a multiplicative, low-frequency bias field, the method evaluates the bias field on a single tissue class, and extends it to the whole image. The algorithm works iteratively, interleaving tissue class domain and bias field estimation using B-spline.