A Modified Reference Scan Method for MR Image Inhomogeneity Correction.

Abstract

Intensity inhomogeneity remains a pivotal challenge that hampers the diagnostic efficacy of Magnetic Resonance Imaging (MRI). Traditional reference scan methods, while effective in correcting intensity inhomogeneity, often inadvertently introduce noise into the images, thus degrading the Signal-to-Noise Ratio (SNR). In this study, we introduce an innovative modified reference scan methodology. Initially, we posit that the sensitivity map utilized for inhomogeneity correction exhibits spatial consistency within the Region of Interest (ROI). Subsequently, we refine the ROI mask extraction process employing a sophisticated clustering method, ensuring adaptability across diverse tissue types. Furthermore, we employ locally weighted smoothing techniques to construct a smooth sensitivity map, effectively precluding the infiltration of noise. Our method was contrasted with other retrospective and prospective correction techniques across phantom, brain, and lumbar MR images. Visual analysis of the results indicates that our method achieves comparable uniformity and superior SNR. Overall, this novel approach substantially enhances MRI quality, thereby augmenting diagnostic precision.