Optimized angle selection for radial sampled NMR experiments

Sparse sampling offers tremendous potential for overcoming the time limitations imposed by traditional Cartesian sampling of indirectly detected dimensions of multidimensional NMR data. Unfortunately, several otherwise appealing implementations are accompanied by spectral artifacts that have the potential to contaminate the spectrum with false peak intensity. In radial sampling of linked time evolution periods, the artifacts are easily identified and removed from the spectrum if a sufficient set of radial sampling angles is employed. Robust implementation of the radial sampling approach therefore requires optimization of the set of radial sampling angles collected. Here we describe several methods for such optimization. The approaches described take advantage of various aspects of the general simultaneous multidimensional Fourier transform in the analysis of multidimensional NMR data. Radially sampled data are primarily contaminated by ridges extending from authentic peaks. Numerical methods are described that definitively identify artifactual intensity and the optimal set of sampling angles necessary to eliminate it under a variety of scenarios. The algorithms are tested with both simulated and experimentally obtained triple resonance data.     

Sampling of the NMR time domain along concentric rings

We present a novel approach to sampling the NMR time domain, whereby the sampling points are aligned on concentric rings, which we term concentric ring sampling (CRS). Radial sampling constitutes a special case of CRS where each ring has the same number of points and the same relative orientation. We derive theoretically that the most efficient CRS approach is to place progressively more points on rings of larger radius, with the number of points growing linearly with the radius, a method that we call linearly increasing CRS (LCRS). For cases of significant undersampling to reduce measurement time, a randomized LCRS (RLCRS) is also described. A theoretical treatment of these approaches is provided, including an assessment of artifacts and sensitivity. The analytical treatment of sensitivity also addresses the sensitivity of radially sampled data processed by Fourier transform. Optimized CRS approaches are found to produce artifact-free spectra of the same resolution as Cartesian sampling, for the same measurement time. Additionally, optimized approaches consistently yield fewer and smaller artifacts than radial sampling, and have a sensitivity equal to Cartesian and better than radial sampling. We demonstrate the method using numerical simulations, as well as a 3D HNCO experiment on protein G B1 domain.     

Polar Fourier transforms of radially sampled NMR data

Radial sampling of the NMR time domain has recently been introduced to speed up data collection significantly. Here, we show that radially sampled data can be processed directly using Fourier transforms in polar coordinates. We present a comprehensive theoretical analysis of the discrete polar Fourier transform, and derive the consequences of its application to radially sampled data using linear response theory. With adequate sampling, the resulting spectrum using a polar Fourier transform is indistinguishable from conventionally processed spectra with Cartesian sampling. In the case of undersampling in azimuth--the condition that provides significant savings in measurement time-the correct spectrum is still produced, but with limited distortion of the baseline away from the peaks, taking the form of a summation of high-order Bessel functions. Finally, we describe an intrinsic connection between the polar Fourier transform and the filtered backprojection method that has recently been introduced to process projection-reconstruction NOESY data. Direct polar Fourier transformation holds great potential for producing quantitatively accurate spectra from radially sampled NMR data.     

基于低秩矩阵的非均匀采样NMR波谱重建进展

多维核磁共振（Nuclear Magnetic Resonance，NMR）利用多维波谱来分析分子结构，被广泛用于化学、生物学和医学等领域，但信号采样时间随波谱维度和采样点数增加而迅速增长．非均匀采样通过降低间接维采样点数来加速数据采集，并引入合理的重建方法获得完整的NMR波谱．如何快速重建高质量的波谱，是NMR信号处理研究的前沿．本文主要综述近年来基于低秩矩阵的NMR波谱重建方法的发展．首先介绍了低秩矩阵的相关数学基础；然后从一般低秩矩阵和结构化低秩汉克尔矩阵两个角度来论述重建模型，并讨论相关的NMR波谱应用；最后分析了该技术存在的不足，并展望其未来发展的趋势．     

Phasing arbitrarily sampled multidimensional NMR data

The recent re-introduction of the two-dimensional Fourier transformation (2D-FT) has allows for the transformation of arbitrarily sampled time domain signals. In this respect, radial sampling, where two incremented time dimensions (t(1) and t(2)) are sampled such that t(1)=taucosalpha and t(2)=tausinalpha, is especially appealing because of the relatively small leakage artifacts that occur upon Fourier transformation. Unfortunately radially sampled time domain data results in a fundamental artifact in the frequency domain manifested as a ridge of intensity extending through the peak positions perpendicular to +/- the radial sampling angle. Successful removal of the ridge artifacts using existing algorithms requires absorptive line shapes. Here we present two procedures for retrospective phase correction of arbitrarily sampled data.     

Application of Maximum Entropy reconstruction to PISEMA spectra

Maximum Entropy reconstruction is applied to two-dimensional PISEMA spectra of stationary samples of peptide crystals and proteins in magnetically aligned virus particles and membrane bilayers. Improvements in signal-to-noise ratios were observed with minimal distortion of the spectra when Maximum Entropy reconstruction was applied to non-linearly sampled data in the indirect dimension. Maximum Entropy reconstruction was also applied in the direct dimension by selecting sub-sets of data from the free induction decays. Because the noise is uncorrelated in the spectra obtained by Maximum Entropy reconstruction of data with different non-linear sampling schedules, it is possible to improve the signal-to-noise ratios by co-addition of multiple spectra derived from one experimental data set. The combined application of Maximum Entropy to data in the indirect and direct dimensions has the potential to lead to substantial reductions in the total amount of experimental time required for acquisition of data in multidimensional NMR experiments.     

Two-dimensional Fourier transform of arbitrarily sampled NMR data sets

A new procedure for Fourier transform with respect to more than one time variable simultaneously is proposed for NMR data processing. In the case of two-dimensional transform the spectrum is calculated for pairs of frequencies, instead of conventional sequence of one-dimensional transforms. Therefore, it enables one to Fourier transform arbitrarily sampled time domain and thus allows for analysis of high dimensionality spectra acquired in a short time. The proposed method is not limited to radial sampling, it requires only to fulfill the Nyquist theorem considering two or more time domains at the same time. We show the application of new approach to the 3D HNCO spectrum acquired for protein sample with radial and spiral time domain sampling.     

Technique for Importing Greater Evolution Resolution in Multidimensional NMR Spectrum

A very simple and general procedure that extracts constant-evolution-frequency data from a truncated multidimensional (2D, 3D, 4D, etc.) FID is described, generalized, analyzed, and illustrated. The method replaces Fourier transformation of the evolution dimension with a linear model created from a separate, high-quality 1D FID. The equivalent of high resolution in the evolution dimension can be achieved without obtaining an extensive multidimensional FID. The analysis of the 1D FID can also be used to predict the signal to noise ratio of the extracted slices that will result from various evolution dimension sampling protocols, making it possible to developa priorian optimal sampling strategy for the multidimensional FID. The evolution dimension need not be sampled periodically. The procedure has a potential signal-to-noise ratio advantage because it extracts usable information from a multidimensional FID at short evolution times before the magnetization has decayed significantly.     

Randomization improves sparse sampling in multidimensional NMR

While a number of strategies have been developed to reduce data collection requirements for multidimensional NMR based on non-Fourier methods of spectrum analysis, there is an increasing awareness that the principal differences in the performance of these methods is attributable to the sampling strategies employed, and not the method of spectrum analysis per se. The ability of maximum entropy reconstruction to utilize essentially arbitrary sampling schemes makes it a useful platform for comparative analysis of sampling strategies. Here we use maximum entropy reconstruction to demonstrate that artifacts characteristic of sparse sampling result from regularity in the sampling pattern, and that they can be substantially reduced by introducing a degree of randomness to an otherwise regular sampling scheme, without requiring additional sampling.     

Fast multidimensional NMR: radial sampling of evolution space

Multidimensional NMR spectroscopy can be speeded up by limited radial sampling of the time-domain evolution data. The resulting frequency-domain projections are used to reconstruct the full NMR spectrum. New algorithms are proposed to suppress back-projection artifacts while retaining optimum sensitivity. The method is illustrated by experiments on the 900 MHz HNCO spectrum of a protein, HasA.     

Ultra-high resolution 3D NMR spectra from limited-size data sets

The advantage of the filter diagonalization method (FDM) for analysis of triple-resonance NMR experiments is demonstrated by application to a 3D constant time (CT) HNCO experiment. With a 15N-,13C-labeled human ubiquitin sample (1.0 mM), high spectral resolution was obtained at 500 MHz in 25 min with only 6-8 increments in each of the CT dimensions. This data set size is about a factor of 50-100 smaller than typically required, yet FDM analysis results in a fully resolved spectrum with a sharp peak for each HNCO resonance. Unlike Fourier transform (FT) processing, in which spectral resolution in each dimension is inversely proportional to the acquisition time in this dimension, FDM is a true multi-dimensional method; the resolution in all dimensions is determined by the total information content of the entire signal. As the CT dimensions of the 3D HNCO signal have approximate time-reversal symmetry, they can each be doubled by combining the usual four hyper-complex data sets. This apparent quadrupling of the data is important to the success of the method. Thus, whenever raw sensitivity is not limiting, well-resolved n-dimensional spectra can now be obtained in a small fraction of the usual time. Alternatively, to maximize sensitivity, evolution periods of faster relaxing nuclei may be radically shortened, the total required resolution being obtained through chemical shift encoding of other, more slowly relaxing, spins. Improvements similar to those illustrated with a 3D HNCO spectrum are expected for other triple-resonance spectra, where CT evolution in the indirect dimensions is implemented.     

Methods of reconstruction of spectra in multidimensional NMR spectroscopy

In this review, some methods for speeding up the performance of multidimensional nuclear magnetic resonance (NMR) experiments are discussed. It is shown that, at a sufficiently high spectral sensitivity, which does not require multiple scanning with averaging, two-dimensional proton-correlation experiments (COSY and TOCSY) can be performed in less than one minute. A multifold decrease in the time of multidimensional experiments can be achieved by various methods, for example, by the direct excitation of resonance signals with a set of different frequencies obtained using the Hadamard matrix. Methods for reconstructing multidimensional NMR spectra based on the inverse Radon transform and a number of other promising methods are also considered     

Spectral estimation of NMR relaxation

In this paper, spectral estimation of NMR relaxation is constructed as an extension of Fourier Transform (FT) theory as it is practiced in NMR or MRI, where multidimensional FT theory is used. nD NMR strives to separate overlapping resonances, so the treatment given here deals primarily with monoexponential decay. In the domain of real error, it is shown how optimal estimation based on prior knowledge can be derived. Assuming small Gaussian error, the estimation variance and bias are derived. Minimum bias and minimum variance are shown to be contradictory experimental design objectives. The analytical continuation of spectral estimation is constructed in an optimal manner. An important property of spectral estimation is that it is phase invariant. Hence, hypercomplex data storage is unnecessary. It is shown that, under reasonable assumptions, spectral estimation is unbiased in the context of complex error and its variance is reduced because the modulus of the whole signal is used. Because of phase invariance, the labor of phasing and any error due to imperfect phase can be avoided. A comparison of spectral estimation with nonlinear least squares (NLS) estimation is made analytically and with numerical examples. Compared to conventional sampling for NLS estimation, spectral estimation would typically provide estimation values of comparable precision in one-quarter to one-tenth of the spectrometer time when S/N is high. When S/N is low, the time saved can be used for signal averaging at the sampled points to give better precision. NLS typically provides one estimate at a time, whereas spectral estimation is inherently parallel. The frequency dimensions of conventional nD FT NMR may be denoted D₁, D₂, etc. As an extension of nD FT NMR, one can view spectral estimation of NMR relaxation as an extension into the zeroth dimension. In nD NMR, the information content of a spectrum can be extracted as a set of n-tuples (ω₁, … ω_n), corresponding to the peak maxima. Spectral estimation of NMR relaxation allows this information content to be extended to a set of (n + 1)-tuples (λ, ω₁, … ω_n), where λ is the relaxation rate.     

Dynamic range improvement in NMR imaging using phase scrambling

Data collection efficiency in NMR imaging is impaired if the dynamic range of the receiver system is limited in comparison with that of the observed signal. This situation may occur in high-resolution proton imaging of large objects at high magnetic field strengths. The efficiency with which information is received can be increased by reducing the peak amplitude of the spin response by varying the phase distribution of the excited spins. This phase scrambling technique may be implemented using tailored RF excitation or by dephasing using nonlinear magnetic field gradients and can be applied in all dimensions of an acquired data set, providing a significant reduction in the dynamic range requirements of the detection electronics. Experimental results using 2D Fourier imaging have obtained up to 25 dB reduction in peak signal intensities. Image signal-to-noise ratios improved up to a factor of 6, with actual values dependent on experimental conditions. Simulation studies show that computational noise introduced during Fourier transformation is significantly reduced when phase scrambling is employed.     

Cogwheel phase cycling in common triple resonance NMR experiments for the liquid phase

We demonstrate that cogwheel phase cycling, which was previously only used in solid state NMR, can be applied to optimize the efficiency of commonly used pulse sequences in multiple resonance liquid-state biomolecular NMR. In favorable cases the required minimum number of scans can be reduced by more than 80% as compared to a corresponding sequence with nested phase cycles. Since cogwheel phase cycling procedures can be designed for a range of scan numbers, and can be combined with pulsed field gradients, the total experiment time can be adjusted closely to the required signal-to-noise ratio with minimal overhead. Examples are shown for 3D-TROSY-HNCO, 3D-TROSY-HNCACO, and 3D-HACACO experiments on diamagnetic and paramagnetic proteins.     

Composite MR image reconstruction and unaliasing for general trajectories using neural networks

In rapid parallel magnetic resonance imaging, the problem of image reconstruction is challenging. Here, a novel image reconstruction technique for data acquired along any general trajectory in neural network framework, called “Composite Reconstruction And Unaliasing using Neural Networks” (CRAUNN), is proposed. CRAUNN is based on the observation that the nature of aliasing remains unchanged whether the undersampled acquisition contains only low frequencies or includes high frequencies too. Here, the transformation needed to reconstruct the alias-free image from the aliased coil images is learnt, using acquisitions consisting of densely sampled low frequencies. Neural networks are made use of as machine learning tools to learn the transformation, in order to obtain the desired alias-free image for actual acquisitions containing sparsely sampled low as well as high frequencies. CRAUNN operates in the image domain and does not require explicit coil sensitivity estimation. It is also independent of the sampling trajectory used, and could be applied to arbitrary trajectories as well. As a pilot trial, the technique is first applied to Cartesian trajectory-sampled data. Experiments performed using radial and spiral trajectories on real and synthetic data, illustrate the performance of the method. The reconstruction errors depend on the acceleration factor as well as the sampling trajectory. It is found that higher acceleration factors can be obtained when radial trajectories are used. Comparisons against existing techniques are presented. CRAUNN has been found to perform on par with the state-of-the-art techniques. Acceleration factors of up to 4, 6 and 4 are achieved in Cartesian, radial and spiral cases, respectively.     

Fast multi-dimensional NMR by minimal sampling

A new scheme is proposed for very fast acquisition of three-dimensional NMR spectra based on minimal sampling, instead of the customary step-wise exploration of all of evolution space. The method relies on prior experiments to determine accurate values for the evolving frequencies and intensities from the two-dimensional 'first planes' recorded by setting t1=0 or t2=0. With this prior knowledge, the entire three-dimensional spectrum can be reconstructed by an additional measurement of the response at a single location (t1( *),t2( *)) where t1( *) and t2( *) are fixed values of the evolution times. A key feature is the ability to resolve problems of overlap in the acquisition dimension. Applied to a small protein, agitoxin, the three-dimensional HNCO spectrum is obtained 35 times faster than systematic Cartesian sampling of the evolution domain. The extension to multi-dimensional spectroscopy is outlined.     

Multiple-quantum magic-angle spinning spectroscopy using nonlinear sampling

NMR spectroscopy is a relatively insensitive technique and many biomolecular applications operate near the limits of sensitivity and resolution. A particularly challenging example is detection of the quadrupolar nucleus 17O, due to its low natural abundance, large quadrupole couplings, and low gyromagnetic ratio. Yet the chemical shift of 17O spans almost 1000 ppm in organic molecules and it serves as a potentially unique reporter of hydrogen bonding in peptides, nucleic acids, and water, and as a valuable complement to 13C and 15N NMR. Recent developments including the multiple-quantum magic-angle spinning (MQMAS) experiment have enabled the detection of 17O in biological solids, but very long data acquisitions are required to achieve sufficient sensitivity and resolution. Here, we perform nonlinear sampling in the indirect dimension of MQMAS experiments to substantially reduce the total acquisition time and improve sensitivity and resolution. Nonlinear sampling prevents the use of the discrete Fourier transform; instead, we employ maximum entropy (MaxEnt) reconstruction. Nonlinearly sampled MQMAS spectra are shown to provide high resolution and sensitivity in several systems, including lithium sulfate monohydrate (LiSO(4)-H(2)17O) and L-asparagine monohydrate (H(2)17O). The combination of nonlinear sampling and MaxEnt reconstruction promises to make the application of 17O MQMAS practical in a wider range of biological systems.     

Theory of mirrored time domain sampling for NMR spectroscopy

A generalized theory is presented for novel mirrored hypercomplex time domain sampling (MHS) of NMR spectra. It is the salient new feature of MHS that two interferograms are acquired with different directionality of time evolution, that is, one is sampled forward from time t=0 to the maximal evolution time tmax, while the second is sampled backward from t=0 to -tmax. The sampling can be accomplished in a (semi) constant time or non constant-time manner. Subsequently, the two interferograms are linearly combined to yield a complex time domain signal. The manifold of MHS schemes considered here is defined by arbitrary settings of sampling phases ('primary phase shifts') and amplitudes of the two interferograms. It is shown that, for any two given primary phase shifts, the addition theorems of trigonometric functions yield the unique linear combination required to form the complex signal. In the framework of clean absorption mode (CAM) acquisition of NMR spectra being devoid of residual dispersive signal components, 'secondary phase shifts' represent time domain phase errors which are to be eliminated. In contrast, such secondary phase shifts may be introduced by experimental design in order to encode additional NMR parameters, a new class of NMR experiments proposed here. For generalization, it is further considered that secondary phase shifts may depend on primary phase shifts and/or sampling directionality. In order to compare with MHS theory, a correspondingly generalized theory is derived for widely used hypercomplex ('States') sampling (HS). With generalized theory it is shown, first, that previously introduced 'canonical' schemes, characterized by primary phases being multiples of π/4, afford maximal intensity of the desired absorptive signals in the absence of secondary phase shifts, and second, how primary phases can be adjusted to maximize the signal intensity provided that the secondary phase shifts are known. Third, it is demonstrated that theory enables one to accurately measure secondary phase shifts and amplitude imbalances. Application to constant time 2D [13C, 1H]-HSQC spectra recorded for a protein sample with canonical MHS/HS schemes showed that accurate CAM data acquisition can be readily implemented on modern spectrometers for experiments based on through-bond polarization transfer. Fourth, when moderate variations of secondary phase shifts with primary phase shift and/or sampling directionality are encountered, generalized theory allowed comparison of the robustness of different MHS/HS schemes for CAM data acquisition, and thus to identify the scheme best suited to suppress dispersive peak components and quadrature image peaks. Moreover, it is shown that for spectra acquired with several indirect evolution periods, the best suited scheme can be identified independently for each of the periods.