Restoration of intensity uniformity of bi-contrast MRI data with bayesian co-occurrence coring

Abstract

The reconstruction in MRI assumes a uniform radiofrequency field. However, this is violated, which leads to anatomically inconsequential intensity non-uniformities. An anatomic region can be imaged with multiple contrasts that result in different non-uniformities. A method is presented for the joint intensity uniformity restoration of two such images. The effect of the intensity distortion on the auto-co-occurrence statistics of each image as well as on the joint-co-occurrence statistics of the two images is modeled. Their non-stationary deconvolution gives Bayesian coring estimates of the images. Further constraints for smoothness, stability, and validity of the non-uniformity estimates are also imposed. The effectiveness and accuracy of the method has been demonstrated extensively with both BrainWeb phantom images as well as with real brain anatomic data of 29 Parkinson’s disease patients.