SNR-enhanced diffusion MRI with structure-preserving low-rank denoising in reproducing kernel Hilbert spaces
Authors:
Gabriel Ramos-Llordén,
Gonzalo Vegas-Sánchez-Ferrero,
Congyu Liao,
Carl-Fredrik Westin,
Kawin Setsompop,
Yogesh Rathi
Abstract:
Purpose: To introduce, develop, and evaluate a novel denoising technique for diffusion MRI that leverages non-linear redundancy in the data to boost the SNR while preserving signal information. Methods: We exploit non-linear redundancy of the dMRI data by means of Kernel Principal Component Analysis (KPCA), a non-linear generalization of PCAto reproducing kernel Hilbert spaces. By map** the sign…
▽ More
Purpose: To introduce, develop, and evaluate a novel denoising technique for diffusion MRI that leverages non-linear redundancy in the data to boost the SNR while preserving signal information. Methods: We exploit non-linear redundancy of the dMRI data by means of Kernel Principal Component Analysis (KPCA), a non-linear generalization of PCAto reproducing kernel Hilbert spaces. By map** the signal to a high-dimensional space, better redundancy is achieved despite nonlinearities in the data thereby enabling better denoising than linear PCA. We implement KPCA with a Gaussian kernel, with parameters automatically selected from knowledge of the noise statistics, and validate it on realistic Monte-Carlo simulations as well as with in-vivo human brain submillimeter resolution dMRI data. We demonstrate KPCA denoising using multi-coil dMRI data also. Results: SNR improvements up to 2.7 X were obtained in real in-vivo datasets denoised with KPCA, in comparison to SNR gains of up to 1.8 X when using state-of-the-art PCA denoising, e.g., Marchenko- Pastur PCA (MPPCA). Compared to gold-standard dataset references created from averaged data, we showed that lower normalized root mean squared error (NRMSE) was achieved with KPCA compared to MPPCA. Statistical analysis of residuals shows that only noise is removed. Improvements in the estimation of diffusion model parameters such as fractional anisotropy, mean diffusivity, and fiber orientation distribution functions (fODFs)were demonstrated. Conclusion:Non-linear redundancy of the dMRI signal can be exploited with KPCA, which allows superior noise reduction/ SNR improvements than state-of-the-art PCA methods, without loss of signal information.
△ Less
Submitted 14 September, 2020;
originally announced September 2020.
