MR signal inhomogeneity correction for visual and computerized atherosclerosis lesion assessment

We are characterizing atherosclerotic disease in patients and animal models using multiple MR images having different contrasts. We use intravascular and surface array coils giving high signal-to-noise but significant sensitivity inhomogeneity. In human carotid images, bias was corrected using a modified adaptive fuzzy c-mean method with a mechanical membrane model of the bias field. Noise reduction filtering, background segmentation, outlier class identification, and signal normalization were all designed to address specific, significant technical issues. In a synthetic image having a bias field measured from our MR system, variations across an area comparable to a carotid artery were reduced from 60% to <5% with processing while the misclassification rate was kept below 4% even with poor SNR (<7). Human carotid images were qualitatively improved and large regions of skeletal muscle were flattened and normalized for inter- and intra-subject variation. The method should facilitate interpretation of artery gray scales for manual plaque characterization and enable computerized plaque classification.