Robust discrimination of glioblastomas from metastatic brain tumors on the basis of single-voxel (1)H MRS

This article investigates methods for the accurate and robust differentiation of metastases from glioblastomas on the basis of single‐voxel ¹H MRS information. Single‐voxel ¹H MR spectra from a total of 109 patients (78 glioblastomas and 31 metastases) from the multicenter, international INTERPRET database, plus a test set of 40 patients (30 glioblastomas and 10 metastases) from three different centers in the Barcelona (Spain) metropolitan area, were analyzed using a robust method for feature (spectral frequency) selection coupled with a linear‐in‐the‐parameters single‐layer perceptron classifier. For the test set, a parsimonious selection of five frequencies yielded an area under the receiver operating characteristic curve of 0.86, and an area under the convex hull of the receiver operating characteristic curve of 0.91. Moreover, these accurate results for the discrimination between glioblastomas and metastases were obtained using a small number of frequencies that are amenable to metabolic interpretation, which should ease their use as diagnostic markers. Importantly, the prediction can be expressed as a simple formula based on a linear combination of these frequencies. As a result, new cases could be straightforwardly predicted by integrating this formula into a computer‐based medical decision support system. This work also shows that the combination of spectra acquired at different TEs (short TE, 20–32 ms; long TE, 135–144 ms) is key to the successful discrimination between glioblastomas and metastases from single‐voxel ¹H MRS. Copyright © 2011 John Wiley & Sons, Ltd.     

The effect of combining two echo times in automatic brain tumor classification by MRS

(1)H MRS is becoming an accurate, non-invasive technique for initial examination of brain masses. We investigated if the combination of single-voxel (1)H MRS at 1.5 T at two different (TEs), short TE (PRESS or STEAM, 20-32 ms) and long TE (PRESS, 135-136 ms), improves the classification of brain tumors over using only one echo TE. A clinically validated dataset of 50 low-grade meningiomas, 105 aggressive tumors (glioblastoma and metastasis), and 30 low-grade glial tumors (astrocytomas grade II, oligodendrogliomas and oligoastrocytomas) was used to fit predictive models based on the combination of features from short-TEs and long-TE spectra. A new approach that combines the two consecutively was used to produce a single data vector from which relevant features of the two TE spectra could be extracted by means of three algorithms: stepwise, reliefF, and principal components analysis. Least squares support vector machines and linear discriminant analysis were applied to fit the pairwise and multiclass classifiers, respectively. Significant differences in performance were found when short-TE, long-TE or both spectra combined were used as input. In our dataset, to discriminate meningiomas, the combination of the two TE acquisitions produced optimal performance. To discriminate aggressive tumors from low-grade glial tumours, the use of short-TE acquisition alone was preferable. The classifier development strategy used here lends itself to automated learning and test performance processes, which may be of use for future web-based multicentric classifier development studies.     

基于多尺度功能脑网络融合特征的抑郁症分类算法

【摘要】提出一种多尺度功能脑网络融合特征的抑郁症分类方法，具体思想包括:首先通过精细化脑区，建立4种不同尺度的脑网络；然后对每种尺度的脑网络分别提取局部特征和全局特征，并将多种尺度脑网络的特征进行有效融合并降维；最后使用支持向量机对患者脑部功能磁共振影像进行分类。试验结果表明，分别提取局部特征和全局特征，并进行有效融合，可以提升识别效果；空间尺度减小会得到更多有效特征，进而能够有效提升分类结果；多尺度特征融合也可以在很大程度上对分类结果起到积极作用。与传统单一大尺度脑网络方法相比，本研究提出的方法获得了更加优秀的试验结果，识别率可达88.67%，充分验证了本研究提出方法的有效性和可行性，并为抑郁症患者的临床诊断与治疗提供生物学依据。     

Regional neonatal brain absolute thermometry by 1H MRS

Therapeutic hypothermia is standard care for infants with moderate to severe encephalopathy. ¹H MRS thermometry (MRSt) measures regional brain absolute temperature using the temperature‐dependent water chemical shift. This study evaluates the clinical feasibility of MRSt in human neonates, and correlates white matter (WM) and thalamus (Thal) MRSt with conventional rectal temperature (T_rectal) measurement. Fifty‐six infants born at term underwent perinatal MRSt for suspected hypoxic–ischaemic brain injury and 33 infants born preterm had MRSt at a term‐equivalent age; 56 of the 89 had T_rectal measured after MRSt of either a Thal or posterior WM voxel, or both. MRSt used point‐resolved spectroscopy (no water suppression; TR = 1370 ms; TE = 288 ms; 1.5 × 1.5 × 1.5 cm³ Thal and 1.1 × 1.3 × 1.4 cm³ WM voxels). Time domain data were phase and frequency corrected before summation and motion‐corrupted data were excluded from further analysis using simple criteria [preprocessing + quality assurance (QA)]. Two published water temperature‐dependence calibrations [both using cerebral creatine (Cr), choline (Cho) and N‐acetylaspartate (Naa) as independent reference peaks] were compared. The temperature measurements derived from Cr, Cho and Naa were combined to give a single amplitude‐weighted combination temperature (T_AWC). WM and Thal T_AWC correlated linearly with T_rectal (Thal slope, 0.82 ± 0.04, R² = 0.85, p < 0.05; WM slope, 0.95 ± 0.04, R² = 0.78, p < 0.05). Preprocessing + QA improved the correlation between WM T_AWC and T_rectal (R² increased from 0.27 to 0.78, p < 0.001). Both calibration datasets showed specific inconsistencies between the temperatures calculated using Cr, Cho and Naa reference peaks when applied to this neonatal dataset. Neonatal MRSt is clinically feasible. Preprocessing + QA improved MRSt reliability in WM. The consideration of MRSt calibration internal biases is necessary before combining MRSt temperatures from multiple reference peaks to obtain T_AWC. Copyright © 2012 John Wiley & Sons, Ltd.     

Improved reproducibility of MRS regional brain thermometry by 'amplitude-weighted combination'

Brain temperature is important in stroke and trauma. In birth asphyxia, hypothermia improves outcome, but local brain temperature information is needed to optimise therapy. The proton MRS water chemical shift (δ(water) ) is temperature dependent, and the in vivo brain temperature has often been estimated by measuring δ(water) relative to the N-acetylaspartate (NAA) singlet methyl resonance. However, the NAA peak amplitude may be reduced if cerebrospinal fluid occupies part of the MRS voxel and because of the lower concentration in immaturity, pathology and neonatal white matter. These factors can increase random and systematic δ(NAA) errors and also, therefore, MRS brain temperature errors. The aim of this study was to improve MRS brain temperature reproducibility and resilience to pathological, developmental and regional peak amplitude variations by amplitude-weighted combination (AWC) of brain temperatures (T(Cho) , T(Cr) and T(NAA) ) determined using the prominent choline (Cho), total creatine (Cr) and NAA resonances separately as chemical shift references. δ(water) - δ(Cho) , δ(water) - δ(Cr) and δ(water) - δ(NAA) were calibrated against tympanic temperature in piglet brain at 7 T (2.5-cm-diameter surface coil over the parietal lobes; binomial water suppression spin-echo sequence; TE = 540 ms; TR = 5 s). Eight normal human infants underwent thalamic region (Thal) and five occipito-parietal (OP) cerebral MRS at 2.4 T [point-resolved spectroscopy (PRESS) localisation; cubic voxel, 8 mL; water suppression off; TE = 270 ms; TR = 2 s]. AWC with T(Cho) , T(Cr) and T(NAA) weighted by the squared Cho, Cr and NAA peak amplitudes provided the smallest intersubject standard deviations: Thal, 0.45°C; OP, 0.33°C (for T(NAA) values of 0.65°C and 1.12°C, respectively). AWC provided resilience against simulated pathological alterations in Cho, Cr and NAA peak amplitudes, with Thal and OP T(AWC) changing by less than 0.04°C. AWC improves both intersubject reproducibility of MRS temperature estimation and resilience against pathological, anatomical and developmental variation of Cho, Cr and NAA peak amplitudes.     

MRS water resonance frequency in childhood brain tumours: a novel potential biomarker of temperature and tumour environment

¹H MRS thermometry has been investigated for brain trauma and hypothermia monitoring applications but has not been explored in brain tumours. The proton resonance frequency (PRF) of water is dependent on temperature but is also influenced by microenvironment factors, such as fast proton exchange with macromolecules, ionic concentration and magnetic susceptibility. ¹H MRS has been utilized for brain tumour diagnostic and prognostic purposes in children; however, the water PRF measure may provide complementary information to further improve characterization. Water PRF values were investigated from a repository of MRS data acquired from childhood brain tumours and children with apparently normal brains. The cohort consisted of histologically proven glioma (22), medulloblastoma (19) and control groups (28, MRS in both the basal ganglia and parietal white matter regions). All data were acquired at 1.5 T using a short T_E (30 ms) single voxel spectroscopy (PRESS) protocol. Water PRF values were calculated using methyl creatine and total choline. Spectral peak amplitude weighted averaging was used to improve the accuracy of the measurements. Mean PRF values were significantly larger for medulloblastoma compared with glioma, with a difference in the means of 0.0147 ppm (p < 0.05), while the mean PRF for glioma was significantly lower than for the healthy cohort, with a difference in the means of 0.0061 ppm (p < 0.05). This would suggest the apparent temperature of the glioma group was ~1.5 °C higher than the medulloblastomas and ~0.7 °C higher than a healthy brain. However, the PRF shift may not reflect a change in temperature, given that alterations in protein content, microstructure and ionic concentration contribute to PRF shifts. Measurement of these effects could also be used as a supplementary biomarker, and further investigation is required. This study has shown that the water PRF value has the potential to be used for characterizing childhood brain tumours, which has not been reported previously. © 2014 The Authors. NMR in Biomedicine published by John Wiley & Sons, Ltd.     

On the relevance of automatically selected single-voxel MRS and multimodal MRI and MRSI features for brain tumour differentiation

In order to evaluate the relevance of magnetic resonance (MR) features selected by automatic feature selection techniques to build classifiers for differential diagnosis and tissue segmentation two data sets containing MR spectroscopy data from patients with brain tumours were investigated. The automatically selected features were evaluated using literature and clinical experience. It was observed that a significant part of the automatically selected features correspond to what is known from the literature and clinical experience. We conclude that automatic feature selection is a useful tool to obtain relevant and possibly interesting features, but evaluation of the obtained features remains necessary.     

Global brain metabolic quantification with whole‐head proton MRS at 3 T

Total N‐acetyl‐aspartate + N‐acetyl‐aspartate–glutamate (NAA), total creatine (Cr) and total choline (Cho) proton MRS (¹H–MRS) signals are often used as surrogate markers in diffuse neurological pathologies, but spatial coverage of this methodology is limited to 1%–65% of the brain. Here we wish to demonstrate that non‐localized, whole‐head (WH) ¹H–MRS captures just the brain's contribution to the Cho and Cr signals, ignoring all other compartments. Towards this end, 27 young healthy adults (18 men, 9 women), 29.9 ± 8.5 years old, were recruited and underwent T₁‐weighted MRI for tissue segmentation, non‐localizing, approximately 3 min WH ¹H–MRS (T_E/T_R/T_I = 5/10 1 /940 ms) and 30 min ¹H–MR spectroscopic imaging (MRSI) (T_E/T_R = 35/2100 ms) in a 360 cm³ volume of interest (VOI) at the brain's center. The VOI absolute NAA, Cr and Cho concentrations, 7.7 ± 0.5, 5.5 ± 0.4 and 1.3 ± 0.2 mM, were all within 10% of the WH: 8.6 ± 1.1, 6.0 ± 1.0 and 1.3 ± 0.2 mM. The mean NAA/Cr and NAA/Cho ratios in the WH were only slightly higher than the “brain‐only” VOI: 1.5 versus 1.4 (7%) and 6.6 versus 5.9 (11%); Cho/Cr were not different. The brain/WH volume ratio was 0.31 ± 0.03 (brain ≈ 30% of WH volume). Air‐tissue susceptibility‐driven local magnetic field changes going from the brain outwards showed sharp gradients of more than 100 Hz/cm (1 ppm/cm), explaining the skull's Cr and Cho signal losses through resonance shifts, line broadening and destructive interference. The similarity of non‐localized WH and localized VOI NAA, Cr and Cho concentrations and their ratios suggests that their signals originate predominantly from the brain. Therefore, the fast, comprehensive WH‐¹H‐MRS method may facilitate quantification of these metabolites, which are common surrogate markers in neurological disorders.     

High-field MRS of the human brain at short TE and TR

In vivo MRS of the human brain at 7 tesla allows identification of a large number of metabolites at higher spatial resolutions than currently possible at lower field strengths. However, several challenges complicate in vivo localization and artifact suppression in MRS at high spatial resolution within a clinically feasible scan time at 7 tesla. Published MRS sequences at 7 tesla suffer from long echo times, inherent signal‐to‐noise ratio (SNR) loss, large chemical shift displacement artifacts or long repetition times because of excessive radiofrequency (RF) power deposition. In the present study a pulse‐acquire sequence was used that does not suffer from these high field drawbacks. A slice selective excitation combined with high resolution chemical shift imaging for in‐plane localization was used to limit chemical shift displacement artifacts. The pulse‐acquire approach resulted in a very short echo time of 1.4 ms. A cost function guided shimming algorithm was developed to constrain frequency offsets in the excited slice, therefore adiabatic frequency selective suppression could be employed to minimize artifacts from high intensity lipids and water signals in the excited slice. The high sensitivity at a TR of 1 s was demonstrated both on a supraventricular slice as well as in an area very close to the skull in the frontal cortex at a nominal spatial resolution of 0.25 cc within a feasible scan time. Copyright © 2011 John Wiley & Sons, Ltd.     

Classification of brain tumours from MR spectra: the INTERPRET collaboration and its outcomes

The INTERPRET project was a multicentre European collaboration, carried out from 2000 to 2002, which developed a decision‐support system (DSS) for helping neuroradiologists with no experience of MRS to utilize spectroscopic data for the diagnosis and grading of human brain tumours. INTERPRET gathered a large collection of MR spectra of brain tumours and pseudo‐tumoural lesions from seven centres. Consensus acquisition protocols, a standard processing pipeline and strict methods for quality control of the aquired data were put in place. Particular emphasis was placed on ensuring the diagnostic certainty of each case, for which all cases were evaluated by a clinical data validation committee. One outcome of the project is a database of 304 fully validated spectra from brain tumours, pseudotumoural lesions and normal brains, along with their associated images and clinical data, which remains available to the scientific and medical community. The second is the INTERPRET DSS, which has continued to be developed and clinically evaluated since the project ended. We also review here the results of the post‐INTERPRET period. We evaluate the results of the studies with the INTERPRET database by other consortia or research groups. A summary of the clinical evaluations that have been performed on the post‐INTERPRET DSS versions is also presented. Several have shown that diagnostic certainty can be improved for certain tumour types when the INTERPRET DSS is used in conjunction with conventional radiological image interpretation. About 30 papers concerned with the INTERPRET single‐voxel dataset have so far been published. We discuss stengths and weaknesses of the DSS and the lessons learned. Finally we speculate on how the INTERPRET concept might be carried into the future. Copyright © 2015 John Wiley & Sons, Ltd.     

Automated whole‐brain N‐acetylaspartate proton MRS quantification

Concentration of the neuronal marker, N‐acetylaspartate (NAA), a quantitative metric for the health and density of neurons, is currently obtained by integration of the manually defined peak in whole‐head proton (¹H)‐MRS. Our goal was to develop a full spectral modeling approach for the automatic estimation of the whole‐brain NAA concentration (WBNAA) and to compare the performance of this approach with a manual frequency‐range peak integration approach previously employed. MRI and whole‐head ¹H‐MRS from 18 healthy young adults were examined. Non‐localized, whole‐head ¹H‐MRS obtained at 3 T yielded the NAA peak area through both manually defined frequency‐range integration and the new, full spectral simulation. The NAA peak area was converted into an absolute amount with phantom replacement and normalized for brain volume (segmented from T₁‐weighted MRI) to yield WBNAA. A paired‐sample t test was used to compare the means of the WBNAA paradigms and a likelihood ratio test used to compare their coefficients of variation. While the between‐subject WBNAA means were nearly identical (12.8 ± 2.5 mm for integration, 12.8 ± 1.4 mm for spectral modeling), the latter's standard deviation was significantly smaller (by ~50%, p = 0.026). The within‐subject variability was 11.7% (±1.3 mm ) for integration versus 7.0% (±0.8 mm ) for spectral modeling, i.e., a 40% improvement. The (quantifiable) quality of the modeling approach was high, as reflected by Cramer–Rao lower bounds below 0.1% and vanishingly small (experimental ‐ fitted) residuals. Modeling of the whole‐head ¹H‐MRS increases WBNAA quantification reliability by reducing its variability, its susceptibility to operator bias and baseline roll, and by providing quality‐control feedback. Together, these enhance the usefulness of the technique for monitoring the diffuse progression and treatment response of neurological disorders. Copyright © 2014 John Wiley & Sons, Ltd.     

ADAM23 methylation and expression analysis in brain tumors

The ADAMs comprises a family of cell surface proteins with putative roles in cell–cell and/or cell–matrix interactions and in protease activities. In this work, we have examined the expression level and the methylation status of the 5′ upstream region of the adhesion molecule ADAM23 in two brain tumor cell lines (A172 and T98G) as well as in three primary brain tumors (one grade II astrocytoma and two meningiomas) and 15 glioblastoma xenografts. Using bisulfite sequencing we verified that the percentage of methylated dinucleotides is higher in T98G when compared to A172 and that methylation significantly correlates with ADAM23 mRNA and protein expression. However, we were unable to detect methylation and down-regulation of the ADAM23 gene in brain tumors. Together, these results indicate that ADAM23 down-regulation by methylation in brain tumors is a rare event and it may help explain why brain tumor metastases are rarely found elsewhere in the body.     

Proton MRS of early post-natal mouse brain modifications in vivo

NMR provides a non-invasive tool for the phenotypic characterisation of mouse models. The aim of the present study was to apply reliable in vivo MRS techniques for non-invasive investigations of brain development in normal and transgenic mice, by monitoring metabolite concentrations in different brain regions. The conditions of anaesthesia, immobilisation and respiratory monitoring were optimized to carry out in vivo MRS studies in young mice. All the experiments were performed in normal mice, at 9.4 T, applying a point-resolved spectroscopy (PRESS) sequence (TR = 2,000 ms; TE = 130 ms). We obtained reproducible in vivo (1)H NMR spectra of wild-type mouse brains as early as post-natal day 5, which allowed us to follow brain maturation variations from post-natal days 5 to 21. The survival rate of animals was between 66 and 90% at post-natal days 5 and 21, respectively. Developmental changes of metabolite concentrations were measured in three brain regions: the thalamus, a region rich in cell bodies, the olfactory bulb, rich in fibre tracts actively myelinated during brain maturation, and the cerebellum. The voxel size varied from 2 to 8 microL according to the size of the brain structure analysed. The absolute concentrations of the total creatine, taurine, total choline, N-acetylaspartate and of the glutamate/glutamine pool were determined from (1)H NMR spectra obtained in the different brain regions at post-natal day 5, 10, 15 and 21. Variations observed during brain development were in accordance with those previously reported in mice using ex vivo MRS studies, and also in rats and humans in vivo. Possibilities of longitudinal MRS analysis in maturing mice brains provide new perspectives to characterise better the tremendous number of transgenic mutant mice generated with the aim of decrypting the complexity of brain development and neurodegenerative diseases but also to follow the impact of environmental and therapeutic factors.     

Assessment of human brain temperature by 1H MRS during visual stimulation and hypercapnia

Brain temperature is determined by the interplay between the cerebral metabolic rate of oxygen (CMRO2) and cerebral blood flow (CBF). In this study, single-voxel 1H nuclear MRS, with an accuracy of +/-0.2 degrees C for temperature determination, was used at 3 T to measure human brain temperature during visual stimulation (which increases both CBF and CMRO2) and hypercapnia (which increases CBF only). Visual stimulation had no detectable effect on brain temperature in the parenchyma showing blood oxygenation level dependent activation. Hypercapnia, leading to an increase in the end tidal CO2 by 8 +/- 2 mm Hg above the baseline, caused a short-lasting decrease in brain temperature of 0.30 +/- 0.33 degrees C. These results indicate that increased CBF may be a key factor, bringing about a small decrease in brain temperature during brain activation. However, the increase in CBF is not sufficient to lower brain temperature in the presence of a concomitant increase in endogenous heat production.     

1H MRS characterization of neurochemical profiles in orthotopic mouse models of human brain tumors

Glioblastoma (GBM), the most common primary brain tumor, is resistant to currently available treatments. The development of mouse models of human GBM has provided a tool for studying mechanisms involved in tumor initiation and growth as well as a platform for preclinical investigation of new drugs. In this study we used ¹H MR spectroscopy to study the neurochemical profile of a human orthotopic tumor (HOT) mouse model of human GBM. The goal of this study was to evaluate differences in metabolite concentrations in the GBM HOT mice when compared with normal mouse brain in order to determine if MRS could reliably differentiate tumor from normal brain. A T_E =19 ms PRESS sequence at 9.4 T was used for measuring metabolite levels in 12 GBM mice and 8 healthy mice. Levels for 12 metabolites and for lipids/macromolecules at 0.9 ppm and at 1.3 ppm were reliably detected in all mouse spectra. The tumors had significantly lower concentrations of total creatine, GABA, glutamate, total N‐acetylaspartate, aspartate, lipids/macromolecules at 0.9 ppm, and lipids/macromolecules at 1.3 ppm than did the brains of normal mice. The concentrations of glycine and lactate, however, were significantly higher in tumors than in normal brain. Copyright © 2014 John Wiley & Sons, Ltd.     

Improvement of resolution for brain coupled metabolites by optimized (1)H MRS at 7T

Resolution enhancement for glutamate (Glu), glutamine (Gln) and glutathione (GSH) in the human brain by TE‐optimized point‐resolved spectroscopy (PRESS) at 7 T is reported. Sub‐TE dependences of the multiplets of Glu, Gln, GSH, γ‐aminobutyric acid (GABA) and N‐acetylaspartate (NAA) at 2.2–2.6 ppm were investigated with density matrix simulations, incorporating three‐dimensional volume localization. The numerical simulations indicated that the C4‐proton multiplets can be completely separated with (TE₁, TE₂) = (37, 63) ms, as a result of a narrowing of the multiplets and suppression of the NAA 2.5 ppm signal. Phantom experiments reproduced the signal yield and lineshape from simulations within experimental errors. In vivo tests of optimized PRESS were conducted on the prefrontal cortex of six healthy volunteers. In spectral fitting by LCModel, Cramér–Rao lower bounds (CRLBs) of Glu, Gln and GSH were 2 ± 1, 5 ± 1 and 6 ± 2 (mean ± SD), respectively. To evaluate the performance of the optimized PRESS method under identical experimental conditions, stimulated‐echo spectra were acquired with (TE, TM) = (14, 37) and (74, 68) ms. The CRLB of Glu was similar between PRESS and short‐TE stimulated‐echo acquisition mode (STEAM), but the CRLBs of Gln and GSH were lower in PRESS than in both STEAM acquisitions. Copyright © 2010 John Wiley & Sons, Ltd.     

基于锁相值和图论的脑功能网络特征提取方法

目的为了研究大脑运动想象时脑功能网络的状态变化,并区分运动想象任务,本文提出一种基于锁相值和图论的脑功能网络特征提取方法。方法首先获取Mu节律和Beta节律的运动想象脑电信号(motor imagery electroencephalography,MI-EEG),计算任意两导相同节律MI-EEG之间的锁相值;然后分别构建两个节律的脑功能网络,并提取6种全局网络特征参数,对其归一化处理后进行串行融合获得特征向量;最后以支持向量机(support vector machine,SVM)作为分类器,采用10折交叉验证法,在BCI Competition III Data Sets IIIa数据集上对两种运动想象任务进行分类。结果相比于其他脑网络特征提取方法,本文方法获得了较高的识别率,最高识别率和平均识别率分别达到100. 00%和83. 33%。结论从脑功能网络的角度,通过构建Mu节律和Beta节律两个运动节律MI-EEG的脑功能网络,提取多个反映大脑网络整体信息的特征,相对于构建单一运动节律MI-EEG的脑功能网络,提取单个网络特征参数,能够有效改善运动想象任务的识别效果,为MI-EEG信号的特征提取方法提供了一种新的思路。     

Prospective diagnostic performance evaluation of single-voxel 1H MRS for typing and grading of brain tumours

The purpose of this study was to evaluate whether single-voxel (1)H MRS could add useful information to conventional MRI in the preoperative characterisation of the type and grade of brain tumours. MRI and MRS examinations from a prospective cohort of 40 consecutive patients were analysed double blind by radiologists and spectroscopists before the histological diagnosis was known. The spectroscopists had only the MR spectra, whereas the radiologists had both the MR images and basic clinical details (age, sex and presenting symptoms). Then, the radiologists and spectroscopists exchanged their predictions and re-evaluated their initial opinions, taking into account the new evidence. Spectroscopists used four different systems of analysis for (1)H MRS data, and the efficacy of each of these methods was also evaluated. Information extracted from (1)H MRS significantly improved the radiologists' MRI-based characterisation of grade IV tumours (glioblastomas, metastases, medulloblastomas and lymphomas) in the cohort [area under the curve (AUC) in the MRI re-evaluation 0.93 versus AUC in the MRI evaluation 0.85], and also of the less malignant glial tumours (AUC in the MRI re-evaluation 0.93 versus AUC in the MRI evaluation 0.81). One of the MRS analysis systems used, the INTERPRET (International Network for Pattern Recognition of Tumours Using Magnetic Resonance) decision support system, outperformed the others, as well as being better than the MRI evaluation for the characterisation of grade III astrocytomas. Thus, preoperative MRS data improve the radiologists' performance in diagnosing grade IV tumours and, for those of grade II-III, MRS data help them to recognise the glial lineage. Even in cases in which their diagnoses were not improved, the provision of MRS data to the radiologists had no negative influence on their predictions.