Spectrum separation resolves partial-volume effect of MRSI as demonstrated on brain tumor scans

Magnetic resonance spectroscopic imaging (MRSI) is currently used clinically in conjunction with anatomical MRI to assess the presence and extent of brain tumors and to evaluate treatment response. Unfortunately, the clinical utility of MRSI is limited by significant variability of in vivo spectra. Spectral profiles show increased variability because of partial coverage of large voxel volumes, infiltration of normal brain tissue by tumors, innate tumor heterogeneity, and measurement noise. We address these problems directly by quantifying the abundance (i.e. volume fraction) within a voxel for each tissue type instead of the conventional estimation of metabolite concentrations from spectral resonance peaks. This 'spectrum separation' method uses the non-negative matrix factorization algorithm, which simultaneously decomposes the observed spectra of multiple voxels into abundance distributions and constituent spectra. The accuracy of the estimated abundances is validated on phantom data. The presented results on 20 clinical cases of brain tumor show reduced cross-subject variability. This is reflected in improved discrimination between high-grade and low-grade gliomas, which demonstrates the physiological relevance of the extracted spectra. These results show that the proposed spectral analysis method can improve the effectiveness of MRSI as a diagnostic tool.     

Comprehensive processing, display and analysis for in vivo MR spectroscopic imaging

Image reconstruction for magnetic resonance spectroscopic imaging (MRSI) requires specialized spatial and spectral data processing methods and benefits from the use of several sources of prior information that are not commonly available, including MRI-derived tissue segmentation, morphological analysis and spectral characteristics of the observed metabolites. In addition, incorporating information obtained from MRI data can enhance the display of low-resolution metabolite images and multiparametric and regional statistical analysis methods can improve detection of altered metabolite distributions. As a result, full MRSI processing and analysis can involve multiple processing steps and several different data types. In this paper, a processing environment is described that integrates and automates these data processing and analysis functions for imaging of proton metabolite distributions in the normal human brain. The capabilities include normalization of metabolite signal intensities and transformation into a common spatial reference frame, thereby allowing the formation of a database of MR-measured human metabolite values as a function of acquisition, spatial and subject parameters. This development is carried out under the MIDAS project (Metabolite Imaging and Data Analysis System), which provides an integrated set of MRI and MRSI processing functions. It is anticipated that further development and distribution of these capabilities will facilitate more widespread use of MRSI for diagnostic imaging, encourage the development of standardized MRSI acquisition, processing and analysis methods and enable improved mapping of metabolite distributions in the human brain.     

Hierarchical non‐negative matrix factorization (hNMF): a tissue pattern differentiation method for glioblastoma multiforme diagnosis using MRSI

MRSI has shown potential in the diagnosis and prognosis of glioblastoma multiforme (GBM) brain tumors, but its use is limited by difficult data interpretation. When the analyzed MRSI data present more than two tissue patterns, conventional non‐negative matrix factorization (NMF) implementation may lead to a non‐robust estimation. The aim of this article is to introduce an effective approach for the differentiation of GBM tissue patterns using MRSI data. A hierarchical non‐negative matrix factorization (hNMF) method that can blindly separate the most important spectral sources in short‐TE ¹H MRSI data is proposed. This algorithm consists of several levels of NMF, where only two tissue patterns are computed at each level. The method is demonstrated on both simulated and in vivo short‐TE ¹H MRSI data in patients with GBM. For the in vivo study, the accuracy of the recovered spectral sources was validated using expert knowledge. Results show that hNMF is able to accurately estimate the three tissue patterns present in the tumoral and peritumoral area of a GBM, i.e. normal, tumor and necrosis, thus providing additional useful information that can help in the diagnosis of GBM. Moreover, the hNMF results can be displayed as easily interpretable maps showing the contribution of each tissue pattern to each voxel. Copyright © 2012 John Wiley & Sons, Ltd.     

Elliptical localization with pulsed second‐order fields (ECLIPSE) for robust lipid suppression in proton MRSI

Proton MRSI has great clinical potential for metabolic mapping of the healthy and pathological human brain. Unfortunately, the promise has not yet been fully achieved due to numerous technical challenges related to insufficient spectral quality caused by magnetic field inhomogeneity, insufficient RF transmit power and incomplete lipid suppression. Here a robust, novel method for lipid suppression in ¹H MRSI is presented. The method is based on 2D spatial localization of an elliptical region of interest using pulsed second‐order spherical harmonic (SH) magnetic fields. A dedicated, high‐amplitude second‐order SH gradient setup was designed and constructed, containing coils to generate Z2, X2Y2 and XY magnetic fields. Simulations and phantom MRI results are used to demonstrate the principles of the method and illustrate the manifestation of chemical shift displacement. ¹H MRSI on human brain in vivo demonstrates high quality, robust suppression of extracranial lipids. The method allows a wide range of inner or outer volume selection or suppression and should find application in MRSI, reduced‐field‐of‐view MRI and single‐volume MRS.     

Improving the spectral resolution and spectral fitting of 1H MRSI data from human calf muscle by the SPREAD technique

Proton magnetic resonance spectroscopic imaging (¹H MRSI) has been used for the in vivo measurement of intramyocellular lipids (IMCLs) in human calf muscle for almost two decades, but the low spectral resolution between extramyocellular lipids (EMCLs) and IMCLs, partially caused by the magnetic field inhomogeneity, has hindered the accuracy of spectral fitting. The purpose of this paper was to enhance the spectral resolution of ¹H MRSI data from human calf muscle using the SPREAD (spectral resolution amelioration by deconvolution) technique and to assess the influence of improved spectral resolution on the accuracy of spectral fitting and on in vivo measurement of IMCLs. We acquired MRI and ¹H MRSI data from calf muscles of three healthy volunteers. We reconstructed spectral lineshapes of the ¹H MRSI data based on field maps and used the lineshapes to deconvolve the measured MRS spectra, thereby eliminating the line broadening caused by field inhomogeneities and improving the spectral resolution of the ¹H MRSI data. We employed Monte Carlo (MC) simulations with 200 noise realizations to measure the variations of spectral fitting parameters and used an F‐test to evaluate the significance of the differences of the variations between the spectra before SPREAD and after SPREAD. We also used Cramer–Rao lower bounds (CRLBs) to assess the improvements of spectral fitting after SPREAD. The use of SPREAD enhanced the separation between EMCL and IMCL peaks in ¹H MRSI spectra from human calf muscle. MC simulations and F‐tests showed that the use of SPREAD significantly reduced the standard deviations of the estimated IMCL peak areas (p < 10^?8), and the CRLBs were strongly reduced (by ~37%). Copyright © 2014 John Wiley & Sons, Ltd.     

Coil combination of multichannel MRSI data at 7 T: MUSICAL

The goal of this study was to evaluate a new method of combining multi‐channel ¹H MRSI data by direct use of a matching imaging scan as a reference, rather than computing sensitivity maps. Seven healthy volunteers were measured on a 7‐T MR scanner using a head coil with a 32‐channel array coil for receive‐only and a volume coil for receive/transmit. The accuracy of prediction of the phase of the ¹H MRSI data with a fast imaging pre‐scan was investigated with the volume coil. The array coil ¹H MRSI data were combined using matching imaging data as coil combination weights. The signal‐to‐noise ratio (SNR), spectral quality, metabolic map quality and Cramér–Rao lower bounds were then compared with the data obtained by two standard methods, i.e. using sensitivity maps and the first free induction decay (FID) data point. Additional noise decorrelation was performed to further optimize the SNR gain. The new combination method improved significantly the SNR (+29%), overall spectral quality and visual appearance of metabolic maps, and lowered the Cramér–Rao lower bounds (?34%), compared with the combination method based on the first FID data point. The results were similar to those obtained by the combination method using sensitivity maps, but the new method increased the SNR slightly (+1.7%), decreased the algorithm complexity, required no reference coil and pre‐phased all spectra correctly prior to spectral processing. Noise decorrelation further increased the SNR by 13%. The proposed method is a fast, robust and simple way to improve the coil combination in ¹H MRSI of the human brain at 7 T, and could be extended to other ¹H MRSI techniques. © 2013 The Authors. NMR in Biomedicine published by John Wiley & Sons, Ltd.     

Using spatial prior knowledge in the spectral fitting of MRS images

We propose a Bayesian smoothness prior in the spectral fitting of MRS images which can be used in addition to commonly employed prior knowledge. By combining a frequency-domain model for the free induction decay with a Gaussian Markov random field prior, a new optimization objective is derived that encourages smooth parameter maps. Using a particular parameterization of the prior, smooth damping, frequency and phase maps can be obtained whilst preserving sharp spatial features in the amplitude map. A Monte Carlo study based on two sets of simulated data demonstrates that the variance of the estimated parameter maps can be reduced considerably, even below the Cramér-Rao lower bound, when using spatial prior knowledge. Long-TE (1)H MRSI at 1.5 T of a patient with a brain tumor shows that the use of the spatial prior resolves the overlapping peaks of choline and creatine when a single voxel method fails to do so. Improved and detailed metabolic maps can be derived from high-spatial-resolution, short-TE (1)H MRSI at 3 T. Finally, the evaluation of four series of long-TE brain MRSI data with various signal-to-noise ratios shows the general benefit of the proposed approach.     

Fast data acquisition techniques in magnetic resonance spectroscopic imaging

Magnetic resonance spectroscopic imaging (MRSI) is an important technique for assessing the spatial variation of metabolites in vivo. The long scan times in MRSI limit clinical applicability due to patient discomfort, increased costs, motion artifacts, and limited protocol flexibility. Faster acquisition strategies can address these limitations and could potentially facilitate increased adoption of MRSI into routine clinical protocols with minimal addition to the current anatomical and functional acquisition protocols in terms of imaging time. Not surprisingly, a lot of effort has been devoted to the development of faster MRSI techniques that aim to capture the same underlying metabolic information (relative metabolite peak areas and spatial distribution) as obtained by conventional MRSI, in greatly reduced time. The gain in imaging time results, in some cases, in a loss of signal‐to‐noise ratio and/or in spatial and spectral blurring. This review examines the current techniques and advances in fast MRSI in two and three spatial dimensions and their applications. This review categorizes the acceleration techniques according to their strategy for acquisition of the k‐space. Techniques such as fast/turbo‐spin echo MRSI, echo‐planar spectroscopic imaging, and non‐Cartesian MRSI effectively cover the full k‐space in a more efficient manner per T_R. On the other hand, techniques such as parallel imaging and compressed sensing acquire fewer k‐space points and employ advanced reconstruction algorithms to recreate the spatial‐spectral information, which maintains statistical fidelity in test conditions (ie no statistically significant differences on voxel‐wise comparisions) with the fully sampled data. The advantages and limitations of each state‐of‐the‐art technique are reviewed in detail, concluding with a note on future directions and challenges in the field of fast spectroscopic imaging.     

Fast nosological imaging using canonical correlation analysis of brain data obtained by two-dimensional turbo spectroscopic imaging

A new fast and accurate tissue typing technique has recently been successfully applied to prostate MR spectroscopic imaging (MRSI) data. This technique is based on canonical correlation analysis (CCA), a statistical method able to simultaneously exploit the spectral and spatial information characterizing the MRSI data. Here, the performance of CCA is further investigated by using brain data obtained by two-dimensional turbo spectroscopic imaging (2DTSI) from patients affected by glioblastoma. The purpose of this study is to investigate the applicability of CCA when typing tissues of heterogeneous tumors. The performance of CCA is also compared with that of ordinary correlation analysis on simulated as well as in vivo data. The results show that CCA outperforms ordinary correlation analysis in terms of robustness and accuracy.     

快速回波平面磁共振谱成像数据重建算法

传统的相位编码磁共振谱成像采集数据需要很长的时间,使得其在临床上的应用受阻。快速回波平面谱成像(Echoplanarspectroscopicimaging,EPSI)技术采用随时间变化的梯度对谱维和空间维同时进行编码,大大减少了数据采集时间。同时,改进EPSI的读出梯度形式还可以提高‘空间-谱’的分辨率。EPSI数据重建算法比较复杂在t方向先分别对奇偶回波数据进行快速傅立叶变换(FastFouriertranslation,FFT),再利用‘偏移’理论进行组合;在kx方向采用网格化算法将不等间隔采集的数据转换到等间隔的直线网格上,再利用FFT进行图像重建;ky方向是相位编码,不需要转换,直接进行FFT即可。     

Pattern recognition analysis of directional intravoxel incoherent motion MRI in ischemic rodent brains

This study aimed to demonstrate a reliable automatic segmentation method for independently separating reduced diffusion and decreased perfusion areas in ischemic stroke brains using constrained nonnegative matrix factorization (cNMF) pattern recognition in directional intravoxel incoherent motion MRI (IVIM‐MRI). First, the feasibility of cNMF‐based segmentation of IVIM signals was investigated in both simulations and in vivo experiments. The cNMF analysis was independently performed for S₀‐normalized and scaled (by the difference between the maximum and minimum) IVIM signals, respectively. Segmentations of reduced diffusion (from S₀‐normalized IVIM signals) and decreased perfusion (from scaled IVIM signals) areas were performed using the corresponding cNMF pattern weight maps. Second, Monte Carlo simulations were performed for directional IVIM signals to investigate the relationship between the degree of vessel alignment and the direction of the diffusion gradient. Third, directional IVIM‐MRI experiments (x, y and z diffusion‐gradient directions, 20 b values at 7 T) were performed for normal (n = 4), sacrificed (n = 1, no flow) and ischemic stroke models (n = 4, locally reduced flow). The results showed that automatic segmentation of the hypoperfused lesion using cNMF analysis was more accurate than segmentation using the conventional double‐exponential fitting. Consistent with the simulation, the double‐exponential pattern of the IVIM signals was particularly strong in white matter and ventricle regions when the directional flows were aligned with the applied diffusion‐gradient directions. cNMF analysis of directional IVIM signals allowed robust automated segmentation of white matter, ventricle, vascular and hypoperfused regions in the ischemic brain. In conclusion, directional IVIM signals were simultaneously sensitive to diffusion and aligned flow and were particularly useful for automatically segmenting ischemic lesions via cNMF‐based pattern recognition.     

A combined MRI and MRSI based multiclass system for brain tumour recognition using LS-SVMs with class probabilities and feature selection

OBJECTIVE: This study investigates the use of automated pattern recognition methods on magnetic resonance data with the ultimate goal to assist clinicians in the diagnosis of brain tumours. Recently, the combined use of magnetic resonance imaging (MRI) and magnetic resonance spectroscopic imaging (MRSI) has demonstrated to improve the accuracy of classifiers. In this paper we extend previous work that only uses binary classifiers to assess the type and grade of a tumour to a multiclass classification system obtaining class probabilities. The important problem of input feature selection is also addressed. METHODS AND MATERIAL: Least squares support vector machines (LS-SVMs) with radial basis function kernel are applied and compared with linear discriminant analysis (LDA). Both a Bayesian framework and cross-validation are used to infer the parameters of the LS-SVM classifiers. Four different techniques to obtain multiclass probabilities as a measure of accuracy are compared. Four variable selection methods are explored. MRI and MRSI data are selected from the INTERPRET project database. RESULTS: The results illustrate the significantly better performance of automatic relevance determination (ARD), in combination with LS-SVMs in a Bayesian framework and coupling of class probabilities, compared to classical LDA. CONCLUSION: It is demonstrated that binary LS-SVMs can be extended to a multiclass classifier system obtaining class probabilities by Bayesian techniques and pairwise coupling. Feature selection based on ARD further improves the results. This classifier system can be of great help in the diagnosis of brain tumours.     

Automatic quality control in clinical 1H MRSI of brain cancer

MRSI grids frequently show spectra with poor quality, mainly because of the high sensitivity of MRS to field inhomogeneities. These poor quality spectra are prone to quantification and/or interpretation errors that can have a significant impact on the clinical use of spectroscopic data. Therefore, quality control of the spectra should always precede their clinical use. When performed manually, quality assessment of MRSI spectra is not only a tedious and time‐consuming task, but is also affected by human subjectivity. Consequently, automatic, fast and reliable methods for spectral quality assessment are of utmost interest. In this article, we present a new random forest‐based method for automatic quality assessment of ¹H MRSI brain spectra, which uses a new set of MRS signal features. The random forest classifier was trained on spectra from 40 MRSI grids that were classified as acceptable or non‐acceptable by two expert spectroscopists. To account for the effects of intra‐rater reliability, each spectrum was rated for quality three times by each rater. The automatic method classified these spectra with an area under the curve (AUC) of 0.976. Furthermore, in the subset of spectra containing only the cases that were classified every time in the same way by the spectroscopists, an AUC of 0.998 was obtained. Feature importance for the classification was also evaluated. Frequency domain skewness and kurtosis, as well as time domain signal‐to‐noise ratios (SNRs) in the ranges 50–75 ms and 75–100 ms, were the most important features. Given that the method is able to assess a whole MRSI grid faster than a spectroscopist (approximately 3 s versus approximately 3 min), and without loss of accuracy (agreement between classifier trained with just one session and any of the other labelling sessions, 89.88%; agreement between any two labelling sessions, 89.03%), the authors suggest its implementation in the clinical routine. The method presented in this article was implemented in jMRUI's SpectrIm plugin. Copyright © 2016 John Wiley & Sons, Ltd.     

Localized NMR spectroscopy in vivo. Progress and problems

Metabolites in brain and muscle of normal human volunteers have been studied by localized 1H and 31P NMR spectroscopy in vivo. Localization was achieved by means of stimulated echo (STEAM) sequences for both water-suppressed 1H NMR (TE = 20 ms) and 31P NMR (TE = 3 ms). Volumes-of-interest and measuring times selected for brain spectroscopy were 8 mL and 6.5 min for 1H NMR and 125 mL and 13 min for 31P NMR, respectively. General problems relating to spatial localization, spectral resolution, and quantitation of in vivo NMR data are discussed with respect to the nucleus and organs under investigation. They are correlated to studies of tissue extracts obtained at field strengths of 2.35 T (Bruker Biospec) and 7.0 T (Bruker MSL 300). Human studies were performed at 2.0 T on a whole-body research system (Siemens Magnetom).     

磁共振谱成像(MRSI)技术的研究进展

磁共振谱成像 (MRSI)在临床诊断中的作用越来越大。目前 ,最有效的快速谱成像法有回波平面法、螺旋轨迹法和阵列采集法。但是 ,这些 MRSI方法的数据采集时间仍然很长 ,速度有待于进一步提高。抑水抑脂脉冲序列基本定型 ,改进的余地不大。在定量的谱分析方面 ,已经实现分析自动化。谱参数估计方法基本完善 ,但在强基线信号时参数估计方法有待于深入研究。目前 ,磁共振谱的图像重建方法局限在 FFT或网格化后的 FFT,这些方法比较简单、快速 ,但也局限了采样脉冲序列 (采样轨迹 )的大胆设计。期望研究出速度更快的 MRSI数据采集脉冲序列。     

Spatio‐spectral regularization to improve magnetic resonance spectroscopic imaging quantification

Magnetic resonance spectroscopic imaging (MRSI) is a non‐invasive technique able to provide the spatial distribution of relevant biochemical compounds commonly used as biomarkers of disease. Information provided by MRSI can be used as a valuable insight for the diagnosis, treatment and follow‐up of several diseases such as cancer or neurological disorders. Obtaining accurate metabolite concentrations from in vivo MRSI signals is a crucial requirement for the clinical utility of this technique. Despite the numerous publications on the topic, accurate quantification is still a challenging problem due to the low signal‐to‐noise ratio of the data, overlap of spectral lines and the presence of nuisance components. We propose a novel quantification method, which alleviates these limitations by exploiting a spatio‐spectral regularization scheme. In contrast to previous methods, the regularization terms are not expressed directly on the parameters being sought, but on appropriate transformed domains. In order to quantify all signals simultaneously in the MRSI grid, while introducing prior information, a fast proximal optimization algorithm is proposed. Experiments on synthetic MRSI data demonstrate that the error in the estimated metabolite concentrations is reduced by a mean of 41% with the proposed scheme. Results on in vivo brain MRSI data show the benefit of the proposed approach, which is able to fit overlapping peaks correctly and to capture metabolites that are missed by single‐voxel methods due to their lower concentrations. Copyright © 2016 John Wiley & Sons, Ltd.     

Classification of prostatic diseases by means of multivariate analysis on in vivo proton MRSI and DCE‐MRI data

Multivariate analysis has been applied on proton magnetic resonance spectroscopic imaging (¹H‐MRSI) and dynamic contrast enhanced MRI (DCE‐MRI) data of patients with different prostatic diseases such as chronic inflammation, fibrosis and adenocarcinoma. Multivariate analysis offers a global view of the entire range of information coming from both the imaging and spectroscopic side of NMR technology, leading to an integrated picture of the system relying upon the entire metabolic and dynamic profile of the studied samples. In this study, we show how this approach, applied to ¹H‐MRSI/DCE‐MRI results, allows us to differentiate among the various prostatic diseases in a non‐invasive way with a 100% accuracy. These findings suggest that multivariate analysis of ¹H‐MRSI/DCE‐MRI can significantly improve the diagnostic accuracy for these pathological entities. From a more theoretical point of view, the complementation of a single biomarker approach with an integrated picture of the entire metabolic and dynamic profile allows for a more realistic appreciation of pathological entities. Copyright © 2009 John Wiley & Sons, Ltd.     

A phase and frequency alignment protocol for 1H MRSI data of the prostate

(1)H MRSI of the prostate reveals relative metabolite levels that vary according to the presence or absence of tumour, providing a sensitive method for the identification of patients with cancer. Current interpretations of prostate data rely on quantification algorithms that fit model metabolite resonances to individual voxel spectra and calculate relative levels of metabolites, such as choline, creatine, citrate and polyamines. Statistical pattern recognition techniques can potentially improve the detection of prostate cancer, but these analyses are hampered by artefacts and sources of noise in the data, such as variations in phase and frequency of resonances. Phase and frequency variations may arise as a result of spatial field gradients or local physiological conditions affecting the frequency of resonances, in particular those of citrate. Thus, there are unique challenges in developing a peak alignment algorithm for these data. We have developed a frequency and phase correction algorithm for automatic alignment of the resonances in prostate MRSI spectra. We demonstrate, with a simulated dataset, that alignment can be achieved to a phase standard deviation of 0.095 rad and a frequency standard deviation of 0.68 Hz for the citrate resonances. Three parameters were used to assess the improvement in peak alignment in the MRSI data of five patients: the percentage of variance in all MRSI spectra explained by their first principal component; the signal-to-noise ratio of a spectrum formed by taking the median value of the entire set at each spectral point; and the mean cross-correlation between all pairs of spectra. These parameters showed a greater similarity between spectra in all five datasets and the simulated data, demonstrating improved alignment for phase and frequency in these spectra. This peak alignment program is expected to improve pattern recognition significantly, enabling accurate detection and localisation of prostate cancer with MRSI.     

Exploiting spatial information to estimate metabolite levels in two-dimensional MRSI of heterogeneous brain lesions

MRSI provides MR spectra from multiple adjacent voxels within a body volume represented as a two- or three-dimensional matrix, allowing the measurement of the distribution of metabolites over this volume. The spectra of these voxels are usually analyzed one by one, without exploiting their spatial context. In this article, we present an advanced metabolite quantification method for MRSI data, in which the available spatial information is considered. A nonlinear least-squares algorithm is proposed in which prior knowledge is included in the form of proximity constraints on the spectral parameters within a grid and optimized starting values. A penalty term that promotes a spatially smooth spectral parameter map is added to the fitting algorithm. This method is adaptive, in the sense that several sweeps through the grid are performed and each solution may tune some hyperparameters at run-time. Simulation studies of MRSI data showed significantly improved metabolite estimates after the inclusion of spatial information. Improved metabolite maps were also demonstrated by applying the method to in vivo MRSI data. Overlapping peaks or peaks of compounds present at low concentration can be better quantified with the proposed method than with single-voxel approaches. The new approach compares favorably against the multivoxel approach embedded in the well-known quantification software LCModel.