Adaptive kernel regression method for fetal brain magnetic resonance imaging denoising
NI Qian
LIU Ben
HAN Yuanfeng
YU Quangui
CHEN Xiongde
WEN Tiexiang
Abstract：For Rician noise existed in magnetic resonance imaging(MRI)of fetal brain,we designed a kernel regression denoising method.Firstly,classical kernel regression(CKR)was used to acquire gradient information.Then,the covariance matrix representing the local characteristics of the MRI image was constructed by the gradient information and the adaptive kernel regression(AKR)to a-chieve adaptive denoising of MRI data.The quantitative analysis results of MRI data of 9 sets of fetal brain MRI data and 12 sets of a-dult brain MRI data with different Rician noise levels showed that the AKR denoising algorithm could reduce the root mean square error(RMSE)by approximately 28.64%～57.57%,the peak signal-to-noise ratio(PSNR)was increased by approximately 11.67%～45.50%,and the structural similarity index measure(SSIM)was increased by approximately 7.95%～72.50%.The qualitative analysis results of the simulation data and the real MRI of fetal brain indicate that this algorithm can effectively remove the noise in MRI data of fetal and adult brain,and can maintain the global features of the images.
Keywords：Fetal brainMagnetic resonance imagingAdaptive kernel regressionImage denoising
Publication Date：2025-12-30
Online Publishing Date：2026-09-11(First online date of this platform, not the publication date of the document)
Pages：8( 363-370 )