Intensity Non-uniformity Correction in MR Imaging Using Residual Cycle Generative Adversarial Network

Abstract

Purpose: Correcting or reducing the effects of voxel intensity non-uniformity (INU) within a given tissue type is a crucial issue for quantitative MRI image analysis in daily clinical practice. In this study, we present a deep learning-based approach for MRI image INU correction. Method: We developed a residual cycle generative adversarial network (res-cycle GAN), which integrates the residual block concept into a cycle-consistent GAN (cycle-GAN). In cycle-GAN, an inverse transformation was implemented between the INU uncorrected and corrected MRI images to constrain the model through forcing the calculation of both an INU corrected MRI and a synthetic corrected MRI. A fully convolution neural network integrating residual blocks was applied in the generator of cycle-GAN to enhance end-to-end raw MRI to INU corrected MRI transformation. A cohort of 30 abdominal patients with T1-weighted MR INU images and their corrections with a clinically established and commonly used method, namely, N4ITK were used as a pair to evaluate the proposed res-cycle GAN based INU correction algorithm. Quantitatively comparisons were made among the proposed method and other approaches. Result: Our res-cycle GAN based method achieved higher accuracy and better tissue uniformity compared to the other algorithms. Moreover, once the model is well trained, our approach can automatically generate the corrected MR images in a few minutes, eliminating the need for manual setting of parameters. Conclusion: In this study, a deep learning based automatic INU correction method in MRI, namely, res-cycle GAN has been investigated. The results show that learning based methods can achieve promising accuracy, while highly speeding up the correction through avoiding the unintuitive parameter tuning process in N4ITK correction.