M. van der Graaf

Brain tumor classification based on long echo proton MRS signals

There has been a growing research interest in brain tumor classification based on proton magnetic resonance spectroscopy (1H MRS) signals. Four research centers within the EU funded INTERPRET project have acquired a significant number of long echo 1H MRS signals for brain tumor classification. In this paper, we present an objective comparison of several classification techniques applied to the discrimination of four types of brain tumors: meningiomas, glioblastomas, astrocytomas grade II and metastases. Linear and non-linear classifiers are compared: linear discriminant analysis (LDA), support vector machines (SVM) and least squares SVM (LS-SVM) with a linear kernel as linear techniques and LS-SVM with a radial basis function (RBF) kernel as a non-linear technique. Kernel-based methods can perform well in processing high dimensional data. This motivates the inclusion of SVM and LS-SVM in this study. The analysis includes optimal input variable selection, (hyper-) parameter estimation, followed by performance evaluation. The classification performance is evaluated over 200 stratified random samplings of the dataset into training and test sets. Receiver operating characteristic (ROC) curve analysis measures the performance of binary classification, while for multiclass classification, we consider the accuracy as performance measure. Based on the complete magnitude spectra, automated binary classifiers are able to reach an area under the ROC curve (AUC) of more than 0.9 except for the hard case glioblastomas versus metastases. Although, based on the available long echo 1H MRS data, we did not find any statistically significant difference between the performances of LDA and the kernel-based methods, the latter have the strength that no dimensionality reduction is required to obtain such a high performance.

Sensitivity-enhanced 13C MR spectroscopy of the human brain at 3 Tesla†

A new coil design for sensitivity‐enhanced 13C MR spectroscopy (MRS) of the human brain is presented. The design includes a quadrature transmit/receive head coil optimized for 13C MR sensitivity. Loss‐less blocking circuits inside the coil conductors allow this coil to be used inside a homogeneous circularly polarized 1H B1 field for 1H decoupled 13C MRS. A quadrature 1H birdcage coil optimized for minimal local RF heating makes broadband 1H decoupling in the entire human brain possible at 3 Tesla while remaining well within international safety guidelines for RF absorption. Apart from a substantial increase in sensitivity compared to conventional small linear coils, the quadrature 13C coil combined with the quadrature 1H birdcage coil allows efficient cross polarization (CP) in the brain, resulting in an additional 3.5‐fold sensitivity improvement compared to direct 13C measurements without nuclear Overhauser enhancement (NOE) or polarization transfer. Combined with the gain in power efficiency, this setup allows broadband 1H to 13C CP over large areas of the brain. Clear 13C resonances from glutamate (Glu), glutamine (Gln), aspartate (Asp), lactate (Lac), and γ‐aminobutyrate (GABA) carbon spins in the human brain demonstrate the quality of 13C MR spectra obtained in vivo with this coil setup. Magn Reson Med, 2006.

Removal of the outer lines of the citrate multiplet in proton magnetic resonance spectra of the prostatic gland by accurate timing of a point-resolved spectroscopy pulse sequence

Proton MR spectra of a healthy human prostatic gland show a major signal for citrate appearing as an AB-type multiplet. After application of multipulse localization sequences, the outer lines of this multiplet often appear with dispersion line shapes disturbing the baseline and interfering with proper quantification of citrate itself and other nearby resonances. Based upon analytical descriptions of the time evaluation of an AB spin system during a point resolved spectroscopy (PRESS) pulse sequence (90x-τ1,-180y-τ2-180y-t), equations were derived representing the intensity of the absorption and dispersion line shape of the outer lines of the citrate multiplet at the top of echo t=τ2-τ1. From these equations, it was calculated that the outer lines of citrate can be removed almost completely using a PRESS pulse sequence with τ1=11 ms and τ2=60 ms. The theoretical description was confirmed by the almost complete disappearance of the two outer citrate resonances inin vitro andin vivo proton MR spectra acquired with this pulse sequence timing.

Automated correction of unwanted phase jumps in reference signals which corrupt MRSI spectra after eddy current correction

A commonly applied step in the postprocessing of gradient localized proton MR spectroscopy, is correction for eddy current effects using the water signal as a reference. However, this method can degrade some of the metabolite signals, in particular if applied on proton MR spectroscopic imaging data. This artifact arises from the water reference signal in the presence of a second signal which resonates close to the main water resonance. The interference of both resonances will introduce jumps in the phase of the reference time domain signal. Using this phase for eddy current correction will result in a ringing artifact in the frequency domain of the metabolite signal over the whole frequency range. We propose a moving window correction algorithm, which screens the phase of reference signals and removes phase jumps in time domain caused by interference of signals from multiple spin systems. The phase jumps may be abrupt or gradually distributed over several time data points. Because the correction algorithm only corrects time data points which contain phase jumps, the phase is minimally disrupted. Furthermore, the algorithm is automated for large datasets, correcting only those water reference signals which are corrupted. After correction of the corrupted reference signals, normal eddy current correction may be performed. The algorithm is compared with a method which uses a low-pass filter and tested on simulated data as well as on in vivo proton spectroscopic imaging data from a healthy volunteer and from patients with a brain tumor.

N-acetyl resonances in in vivo and in vitro NMR spectroscopy of cystic ovarian tumors.

An unassigned and prominent resonance in the region from δ 2.0–2.1 ppm has frequently been found in the in vivo MR spectra of cancer patients. We demonstrated the presence of this resonance with in vivo MRS in the cyst fluid of a patient with an ovarian tumor. 1H‐NMRS on the aspirated cyst fluid of this patient confirmed the observation. A complex of resonances was observed between 2.0 and 2.1 ppm. It was also present in 11 additional ovarian cyst fluid samples randomly chosen from our biobank. The resonance complex was significantly more prominent in samples from mucinous tumors than in samples from other histological subtypes. A macromolecule (>10 kDa) was found responsible for this complex of resonances. A correlation spectroscopy (COSY) experiment revealed cross peaks of two different types of bound sialic acid suggesting that N‐glycans from glycoproteins and/or glycolipids cause this resonance complex. In the literature, plasma α‐1 acid glycoprotein (AGP), known for its high content of N‐linked glycans, has been suggested to contribute to the δ 2.0–2.1 spectral region. The AGP cyst fluid concentration did not correlate significantly with the peak height of the δ 2.0–2.1 resonance complex in our study. AGP may be partly responsible for the resonance complex but other N‐acetylated glycoproteins and/or glycolipids also contribute. After deproteinization of the cyst fluid, N‐acetyl‐L‐aspartic acid (NAA) was found to contribute significantly to the signal in this spectral region in three of the 12 samples. GC‐MS independently confirmed the presence of NAA in high concentration in the three samples, which all derived from benign serous tumors. We conclude that both NAA and N‐acetyl groups from glycoproteins and/or glycolipids may contribute to the δ 2.0–2.1 ppm resonance complex in ovarian cyst fluid. This spectral region seems to contain resonances from biomarkers that provide relevant clinical information on the type of ovarian tumor. Copyright

Does the combination of magnetic resonance imaging and spectroscopic imaging improve the classification of brain tumours

Magnetic resonance imaging (MRI) and magnetic resonance spectroscopic imaging (MRSI) play an important role in the noninvasive diagnosis of brain tumours. We investigate the use of both MRI and MRSI, separately and in combination with each other for classification of brain tissue types. Many clinically relevant classification problems are considered; for example healthy versus tumour tissues, low- versus high-grade tumours. Linear as well as nonlinear techniques are compared. The classification performance is evaluated by the area under the receiver operating characteristic (ROC) curve (AUC). In general, all techniques achieve a high performance, except when using MRI alone. For example, for low- versus high-grade tumours, low- versus high-grade gliomas, gliomas versus meningiomas, respectively a test AUC higher than 0.91, 0.93 and 0.98 is reached, when both MRI and MRSI data are used.

In vivo proton MR spectroscopy findings specific for adenylosuccinate lyase deficiency.

Adenylosuccinate lyase (ADSL) deficiency is an inherited metabolic disorder affecting predominantly the central nervous system. The disease is characterized by the accumulation of succinylaminoimidazolecarboxamide riboside and succinyladenosine (S‐Ado) in tissue and body fluids. Three children presented with muscular hypotonia, psychomotor delay, behavioral abnormalities, and white matter changes on brain MRI. Two of them were affected by seizures. Screening for inborn errors of metabolism including in vitro high resolution proton MRS revealed an ADSL deficiency that was confirmed genetically in all cases. All patients were studied by in vivo proton MRS. In vitro high resolution proton MRS of patient cerebrospinal fluid showed singlet resonances at 8.27 and 8.29 ppm that correspond to accumulated S‐Ado. In vivo proton MRS measurements also revealed a prominent signal at 8.3 ppm in gray and white matter brain regions of all patients. The resonance was undetectable in healthy human brain. In vivo proton MRS provides a conclusive finding in ADSL deficiency and represents a reliable noninvasive diagnostic tool for this neurometabolic disorder. Copyright

Intake of 13C-4 creatine enables simultaneous assessment of creatine and phosphocreatine pools in human skeletal muscle by 13C MR spectroscopy

The feasibility of a novel method for the noninvasive and local assessment of creatine (Cr) and phosphocreatine (PCr) dynamics in human skeletal muscle based on 13C magnetic resonance (MR) spectroscopy is presented. A high dose of Cr, labeled at the guanidino C‐4 position with 13C 11% enrichment, was administered orally to a human subject for 5 days. Using a surface coil, 13C MR spectra of the lower leg were acquired on a 1.5T MR system at regular time intervals during and after Cr supplementation. An almost twofold increase in the intensities of the resolved PCr and Cr 13C‐4 signals was observed during this period. The slow decrease in these signals to normal values after supplementation reflects the slow daily turnover of Cr. The PCr/Cr ratio did not appear to change over the whole measurement period. During exercise of the leg, reversible changes in PCr and Cr signals were observed, reflecting conversion by the Cr kinase reaction. Magn Reson Med, 2006.

Brain Lactate Concentration Falls in Response to Hypoglycemia in Patients With Type 1 Diabetes and Impaired Awareness of Hypoglycemia

Brain lactate may be involved in the development of impaired awareness of hypoglycemia (IAH), a condition that affects approximately 25% of patients with type 1 diabetes and increases the risk of severe hypoglycemia. The aim of this study was to investigate the effect of acute hypoglycemia on brain lactate concentration in patients with IAH as compared with those with normal awareness of hypoglycemia (NAH) and healthy control subjects (n = 7 per group). After an overnight fast, all subjects underwent a two-step hyperinsulinemic euglycemic (5.0 mmol/L)–hypoglycemic (2.8 mmol/L) glucose clamp. Brain lactate concentrations were measured continuously with 1H-MRS using a specific lactate detection method. Hypoglycemia generated symptoms in patients with NAH and healthy control subjects but not in patients with IAH. Brain lactate fell significantly by ∼20% in response to hypoglycemia in patients with type 1 diabetes with IAH but remained stable in both healthy control subjects and in patients with NAH. The fall in brain lactate is compatible with increased brain lactate oxidation providing an alternative fuel source during hypoglycemia, which may contribute to the impaired detection of hypoglycemia.

Transmit/receive headcoil for optimal 1H MR spectroscopy of the brain in paediatric patients at 3T

Abstract1H magnetic resonance (MR) spectroscopy is a useful tool to obtain metabolic information from the brain in paediatric patients. To detect signals of metabolites at low concentrations or from small volumes, the signal-to-noise ratio (SNR) has to be optimized. The SNR can be increased by going to higher field strengths. However, this leads to higher spectral bandwidths, which increases the chemical shift artefact. Here we present a transmit/receive headcoil which is adapted to the dimensions of the paediatric head and enables PRESS localization with high radio-frequency (RF) bandwidths that minimize the chemical shift displacement to only 5%. In addition, since the pulse lengths are shorter with higher RF bandwidths, the echo time can be reduced to 10 ms improving SNR as well.