PCA denoising and Wiener deconvolution of ³¹P 3D CSI data to enhance effective SNR and improve point spread function

Abstract

Purpose This study evaluates the performance of 2 processing methods, that is, principal component analysis‐based denoising and Wiener deconvolution, to enhance the quality of phosphorus 3D chemical shift imaging data. Methods Principal component analysis‐based denoising increases the SNR while maintaining spectral information. Wiener deconvolution reduces the FWHM of the voxel point spread function, which is increased by Hamming filtering or Hamming‐weighted acquisition. The proposed methods are evaluated using simulated and in vivo 3D phosphorus chemical shift imaging data by 1) visual inspection of the spatial signal distribution; 2) SNR calculation of the PCr peak; and 3) fitting of metabolite basis functions. Results With the optimal order of processing steps, we show that the effective SNR of in vivo phosphorus 3D chemical shift imaging data can be increased. In simulations, we show we can preserve phosphorus‐containing metabolite peaks that had an SNR < 1 before denoising. Furthermore, using Wiener deconvolution, we were able to reduce the FWHM of the voxel point spread function with only partially reintroducing Gibb‐ringing artifacts while maintaining the SNR. After data processing, fitting of the phosphorus‐containing metabolite signals improved. Conclusion In this study, we have shown that principal component analysis‐based denoising in combination with regularized Wiener deconvolution allows increasing the effective spectral SNR of in vivo phosphorus 3D chemical shift imaging data, with reduction of the FWHM of the voxel point spread function. Processing increased the effective SNR by at least threefold compared to Hamming weighted acquired data and minimized voxel bleeding. With these methods, fitting of metabolite amplitudes became more robust with decreased fitting residuals.

Residual quadrupolar couplings observed in 7 Tesla deuterium MR spectra of skeletal muscle

“…Channel combination was performed using the Roemer equal noise algorithm 28 . Spectra were denoised with a principal component analysis–noise suppression algorithm 29,30 and zero‐filled to 2048 points. The linewidths and splittings of the HDO signal in each voxel were determined by fitting the signal with 2 Lorentzian lines, with equal linewidth, amplitude, and phase, using AMARES (Advanced Method for Accurate, Robust, and Efficient Spectral fitting) 31 …”

Increased cardiac Pi/PCr in the diabetic heart observed using phosphorus magnetic resonance spectroscopy at 7T

“…Also, due to the high BMI of several patients, significantly increasing the distance between the coil and the septum, the SNR of the Pi peak was too low to allow reliable Pi/PCr quantification in 5 out of 17 recruited patients. The use of dedicated higher B 1 + performance RF coils, e.g., quadrature [ 32 , 33 ], or a birdcage design [ 34 ] combined with a similarly or more sensitive receive array as used here [ 35 , 36 ] could improve the achievable spectral SNR in such high BMI patients in the future. While we used a dedicated B 0 shimming procedure, additional higher order shimming might improve the field homogeneity further and yield narrower linewidth and thus higher SNR.…”

Uncertainty in denoising of MRSI using low‐rank methods

“…Both SURE and MP automated threshold selection lead to biased data in low SNR cases (because the rank threshold tends to 1 and signal variance is lost). The lower bias and skewness of the SURE SVT (soft thresholding) at high noise levels indicates that it or other thresholding functions may be optimal for low SNR regimes (or where trying to detect small signal fluctuations), such as phosphorus-31 MRS or diffusion-weighted MRS. 9 , 31 In fact, recent work investigating optimal matrix denoising 29 highlights the increased performance of soft thresholding in lower SNR regimes and also proposes optimized singular value shrinkage functions that outperform hard or soft thresholding in all SNR regimes. Evaluation of this parameter-free optimized singular value shrinkage method in both simulation and in vivo data would certainly be an interesting direction for future work.…”

“…High levels of apparent denoising are consistently achieved by denoising algorithms in MRS with additional encoding dimensions, 7 , 8 or MRSI. 1 , 9 , 10 However, it is not clear whether there is an overall reduction in uncertainty of final dynamic model parameters, 11 or metabolite concentrations (the typical output of MRSI). Denoising will lower the apparent noise in any given spectrum, but can introduce systematic model-based errors affecting the bias and variance of the output, ultimately resulting in lower reproducibility and higher mean squared error.…”

“…High levels of apparent denoising are consistently achieved by denoising algorithms in MRS with additional encoding dimensions, 7,8 or MRSI 1,9,10 . However, it is not clear whether there is an overall reduction in uncertainty of final dynamic model parameters, 11 or metabolite concentrations (the typical output of MRSI).…”