Restoration of Bi-Contrast MRI Data for Intensity Uniformity with Bayesian Coring of Co-Occurrence Statistics

Abstract

Abstract:The reconstruction of MRI data assumes a uniform radio-frequency field. However, in practice, the radio-frequency field is inhomogeneous and leads to anatomically inconsequential intensity non-uniformities across an image. An anatomic region can be imaged with multiple contrasts reconstructed independently and be suffering from different non-uniformities. These artifacts can complicate the further automated analysis of the images. 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 and used for their non-stationary restoration followed by their back-projection to the images. Several constraints that ensure a stable restoration are also imposed. Moreover, the method considers the inevitable differences between the signal regions of the two images. The method has been evaluated extensively with BrainWeb phantom brain data as well as with brain anatomic data from the Human Connectome Project (HCP) and with data of Parkinson's disease patients. The performance of the proposed method has been compared with that of the N4ITK tool. The proposed method increases tissues contrast at least 4.62 times more than the N4ITK tool for the BrainWeb images. The dynamic range with the N4ITK method for the same images is increased by up to +29.77%, whereas, for the proposed method, it has a corresponding limited decrease of −1.15%, as expected. The validation has demonstrated the accuracy and stability of the proposed method and hence its ability to reduce the requirements for additional calibration scans.

A Review of Image Processing Techniques Common in Human and Plant Disease Diagnosis

“…These features are used to statistically measure the degree of symmetry between the two brain hemispheres. Bayesian coring of co-occurrence statistics are used for the restoration of bicontrast MRI data for intensity uniformity in [5].…”

“…Brain magnetic resonance imaging (MRI) [5][6][7][8][9][10][11] can be used for the diagnosis of glioma, AIDS dementia, Alzheimer's, cancer metastasis, etc. The image processing is applied to T1-weighted (T1-w) and T2-w MRI scans.…”

Restoration for Intensity Nonuniformities with Discontinuities in Whole-Body MRI

“…The method uses a non-parametric Bayesian coring formulation for intensity restoration. This is applied to the statistical representation of the auto-co-occurrence statistics of each image as well as to the joint-co-occurrence statistics between the two images [49][50][51][52]. The effects of the intensity distortions on both types of co-occurrence statistics are modeled and restored.…”