Multiproject–multicenter evaluation of automatic brain tumor classification by magnetic resonance spectroscopy

Justification  Automatic brain tumor classification by MRS has been under development for more than a decade. Nonetheless, to our knowledge, there are no published evaluations of predictive models with unseen cases that are subsequently acquired in different centers. The multicenter eTUMOUR project (2004–2009), which builds upon previous expertise from the INTERPRET project (2000–2002) has allowed such an evaluation to take place. Materials and Methods  A total of 253 pairwise classifiers for glioblastoma, meningioma, metastasis, and low-grade glial diagnosis were inferred based on 211 SV short TE INTERPRET MR spectra obtained at 1.5 T (PRESS or STEAM, 20–32 ms) and automatically pre-processed. Afterwards, the classifiers were tested with 97 spectra, which were subsequently compiled during eTUMOUR. Results  In our results based on subsequently acquired spectra, accuracies of around 90% were achieved for most of the pairwise discrimination problems. The exception was for the glioblastoma versus metastasis discrimination, which was below 78%. A more clear definition of metastases may be obtained by other approaches, such as MRSI + MRI. Conclusions  The prediction of the tumor type of in-vivo MRS is possible using classifiers developed from previously acquired data, in different hospitals with different instrumentation under the same acquisition protocols. This methodology may find application for assisting in the diagnosis of new brain tumor cases and for the quality control of multicenter MRS databases.     

Quantification of Choline-containing Compounds in Malignant Breast Tumors by 1H MR Spectroscopy Using Water as an Internal Reference at 1.5 T

The quantification of choline-containing compounds (Cho) in breast tumors by proton MR spectroscopy (¹H-MRS) has been of great interest because such compounds have been linked to malignancy. In this study, an internal reference method for the absolute quantification of Cho metabolite in malignant breast tumors was presented using a clinical 1.5 T scanner. We performed in vitro measurements to examine the accuracy of absolute quantification using four phantoms of known choline chloride concentrations. There was a high correlation between the calculated concentrations by MRS and the known concentrations (r ² > 0.98). We applied the technique to in vivo breast study conducted on 45 patients with biopsy-confirmed breast cancer. After T ₁ and T ₂ relaxation times were corrected, the Cho levels in this work had a range of 0.76 – 21.20 mmol/kg from 34 MR spectra of 32 patients with malignant breast lesions. This result was rather consistent with the previously published value (i.e., 1.38 – 10 mmol/kg, Bolan et al. in Magn Reson Med 50:1134–1143, 2003). Therefore, we conclude that the internal method using the fully relaxed water as a reference could be used for quantifying Cho metabolite accurately in breast cancer patients using a clinical 1.5 T scanner.     

Age-dependent MRI-detected lesions at early stages of transient global ischemia in rat brain

Although ischemic stroke has higher incidence and severity in aged than in young humans, the age factor is generally neglected in ischemia animal models. This study was aimed at comparing age-dependent effects at early stages of transient global cerebral ischemia (TGCI) in rats. TGCI was induced in two groups of rats (3–6 and 20–24 months old, respectively) by exposure to 15% oxygen and 15 min occlusion of the two common carotid arteries. Brains were analysed in vivo by MRI–apparent diffusion coefficient (ADC) and T₂ maps–at 1–3 h post-TGCI and in vitro by histochemical examination of triphenyltetrazolium chloride (TTC)-stained slices. At 1–3 h post-TGCI, a higher incidence of lesions was found in aged than in young rats especially in the hippocampus and cortex (occipital plus parietal) but not in the thalamus. The lesioned regions showed lower ADC values in aged than in younger rats. The most substantial ADC decreases were associated with enhanced spin-spin relaxation and lower TTC staining. The different responses of the two age groups support the use of aged animals for investigations on different ischemia models. Our model of brain ischemia appears appropriate for further studies including drug effects.     

The C-neu mammary carcinoma in Oncomice; characterization and monitoring response to treatment with herceptin by magnetic resonance methods

To characterize spontaneously occurring c-neu/HER2 overexpressing tumours in oncomice and their response to herceptin by non-invasive magnetic resonance spectroscopy (MRS) and magnetic resonance imaging (MRI). Oncomice were monitored by localized ³¹P MRS during unperturbed growth and before and after treatment with 10 mg/kg herceptin (Hoffman La Roche) intraperitoneally for up to 21 days post-treatment. Vascular morphology and function was assessed by quantitation of tumour magnetic resonance (MR) relaxation rates R₂* and R₂ prior to and either during carbogen (95% O₂/5% CO₂) breathing or following administration of the blood-pool contrast agent NC100150 (Clariscan, Amersham Health). Immunohistochemistry showed strong membrane staining for HER2 protein overexpression. The ³¹P MRS showed only a significant (p<0.01) increase of phosphomonoester / total phosphate ratio over 21 days of growth. Herceptin increased the tumour volume doubling time compared to untreated tumours and significantly increased the phosphomonoester / -nucleoside triphosphate ratio 2 days after treatment (p=0.01). Tumours showed a highly heterogeneous yet significant (p<0.01) decrease or increase in R₂* in response to carbogen or NC100150 respectively. The absence of a decline in tumour bioenergetics with growth, commonly seen in ³¹P MRS studies of transplanted rodent tumour models, coupled with the heterogeneous blood volume revealed by ¹H MRI, suggest a metabolic and vascular phenotype similar to that found in human tumours.     

Reduced glutamate neurotransmission in patients with Alzheimer's disease–an in vivo <Superscript>13</Superscript>C magnetic resonance spectroscopy study

Cognitive impairment in Alzheimer's disease (AD) is not fully explained. PET indicates reduced cerebral metabolic rate for glucose. Since glutamate neurotransmission (GNT) consumes more than 80% of the ATP generated from metabolism, a pilot study was carried out to determine the neuronal tricarboxylic acid cycle (TCA) based on the hypothesis that reduced GNT could contribute to cognitive impairment in AD. Three AD patients with cognitive impairment (mini-mental state exam: 24 vs 30, P<0.05) and significant reduction in both N-acetyl aspartate (NAA)/Creatine (Cr) (P<0.009) and NAA/myo-inositol (mI) ratio (P<0.01), and three age-matched controls each received 0.014–0.016 g/kg/min 99%1–13C glucose IV. Quantitative ¹H and proton-decoupled ¹³C MR brain spectra were acquired from combined posterior-parietal white matter and posterior-cingulate gray matter every 5 min for 140 min. ¹³C magnetic resonance spectroscopy (MRS) measures of glucose oxidation and neuronal TCA rate, including prolonged time to ¹³C enrichment of glutamate (Glu₂) (P<0.004) and bicarbonate (HCO₃) (P<0.03) as well as reduced relative enrichment of Glu₂/Glu₄ between 60 and 100 min (P<0.04), were significantly different in AD patients vs. controls. ¹³C measures of GNT, glutamine (Gln)₂/Glu₂ (P<0.02) and rates of glutamate enrichment (Glu₂/glucose: 0.34 vs 0.86, P=ns and Glu₄/glucose 0.26 vs 0.83, P=ns), were also reduced. ¹³C MRS measures of neuronal TCA cycle, glucose oxidation and GNT were significantly correlated with measures of neuronal integrity: NAA/Cr, [NAA] and mI/NAA as determined by ¹H MRS (R ²=0.73–0.95; P<0.05–0.01), suggesting that impairment of GNT may be a contributing factor in the cognitive impairment characteristic of AD. Electronic Publication     

Lipid composition changes in normal breast throughout the menstrual cycle

The detection of breast cancer in women using magnetic resonance imaging (MR) is increasingly used as a supplement to X-ray mammography. Furthermore, proton MR spectroscopy (¹HMRS) has detected alterations in lipid profiles that are linked with tumor development and progression in human biopsy tissue. Because normal “resting” breast is highly active, it is necessary to consider that any alterations observed in lipid profiles may not be indicative of breast tumor development. The purpose of this study was to assess the changes in lipid composition in the breast throughout the menstrual cycle in “normals’ using MRS at 4.0 T. Five women with no known history of breast disease were subject to biweekly MRS breast examinations. MRS results showing water and fat resonances revealed cyclic changes in the lipid content throughout the duration of the menstrual cycle. In particular, intensity changes were seen in methylene (-CH₂-) and allylic methylene (CH₂CH₂ ^*CH=) resonances at 2.1 ppm and 1.3 ppm, respectively. These intensity changes assumed a similar cyclic trend for each subject over the 28 days that correlate with the follicular, ovulatory, and luteal phases of the menstrual cycle. The results obtained may indicate cell synthesis or metabolic activity in the breast during the menstrual cycle and provide valuable information pertinent to lipid responses associated with breast disease.     

PRESS-based proton single-voxel spectroscopy and spectroscopic imaging with very short echo times using asymmetric RF pulses

Modified point-resolved spectroscopy (PRESS) sequences for single voxel spetroscopy (MRS) and spectroscopic imaging (SI) with very short echo time (T _E ) are described using asymmetric radio-frequency (RF) pulses as well as an optimized design and timing of the PRESS sequence. The proposed sequences were implemented on a standard 4.7 T imaging system yielding a T _E of 6.0 ms only. Simulations and experimental data measured on phantoms and the rat brain in vivo are presented for MRS and SI showing a high signal-to-noise ratio and hardly any phase distortions caused by J-coupling.     

In vivo quantification of response to treatment in patients with multiple myeloma by <Superscript>1</Superscript>H magnetic resonance spectroscopy of bone marrow

Object Magnetic resonance imaging (MRI) is the gold standard non-invasive technique to detect malignant disease in the bone marrow. Proton magnetic resonance spectroscopy (MRS) can be performed as a quick adjunct to routine spinal MRI. We performed proton MRS to patients with multiple myeloma (MM) at diagnosis and after treatment to investigate the possible correlation of MRS data with response to therapy. Patients and methods Twenty-one patients with newly diagnosed MM underwent combined MRI/MRS explorations of a transverse center section in the fifth lumbar vertebral body. MRS was acquired with STEAM and 40 ms TE. Areas of unsuppressed water and lipid resonances were used to calculate the lipid-to-water ratio (LWR). Results No association was detected between initial LWRs and the clinical characteristics of patients. Post treatment MRS was available in 16 patients of whom 11 (69%) presented an LWR increase, this included all complete responders (8/8, 100%, P = 0.012). A post-treatment LWR value equal to or larger than one is proposed as a non-invasive marker of complete response to treatment. Conclusion Only patients responding to treatment presented a significant increase in bone marrow LWR after therapy. MRS may provide an adequate quantification of response to chemotherapy in patients with MM.     

A Multi-Centre, Web-Accessible and Quality Control-Checked Database of in vivo MR Spectra of Brain Tumour Patients

Objective: To describe an Internet-accessible database that contains validated in vivo MR spectra and clinical data of brain tumour patients. Materials and methods: All data from patients entering the INTERPRET project (International Network for Pattern Recognition of Tumours Using Magnetic Resonance, ) were stored in a web-accessible database (iDB) and selected using its query functionality. Criteria for selection were that the case had a single voxel (SV) short-echo (20–32 ms) 1.5 T spectrum acquired from a nodular region of the tumour, that the voxel had been positioned in the same region as where subsequent biopsy was obtained, that the short-echo spectrum had not been discarded because of acquisition artefacts or other reasons, and that a histopathological diagnosis was agreed among a committee of neuropathologists. When the spectra were obtained from normal volunteers or were of abscesses or clinically proven metastases, biopsy was not required. Results: A subset of 304 cases (22 normal volunteers and 282 tumour patients) was obtained. These cases were migrated to another similar database (validated-DB). Conclusion: The validated-DB complies with ethics regulations and represents the population studied. It is accessible by neuroradiologists willing to use information provided by MRS to help in the non-invasive diagnosis of brain tumours     

In vivo MRS of locally advanced breast cancer: characteristics related to negative or positive choline detection and early monitoring of treatment response

Object  

To explore factors determining the detection of total choline (tCho) by in vivo MR spectroscopy (MRS) in locally advanced breast cancer and to evaluate the ability of in vivo tCho to predict treatment response after one cycle of neoadjuvant chemotherapy (NAC).     

ESMRMB 2016, 33rd Annual Scientific Meeting,Vienna, AT,September 29–October 1: Abstracts,Thursday

Objective

To evaluate the feasibility of in vivo measurement of the fatty acid (FA) composition of breast adipose tissue by MRS on a clinical platform.

Material and methods

MRS experiments were performed at 3 T, using a STEAM sequence, on 25 patients diagnosed with breast cancer. MR spectra, acquired on healthy breast tissue, were analysed with the LCModel.

Results

The measured values of the saturated fatty acid (SFA), mono-unsaturated fatty acid (MUFA) and poly-unsaturated fatty acid (PUFA) fractions were 23.8 ± 7.1 %, 55.4 ± 6.8 % and 20.8 ± 4.4 %, respectively.The values of SFA, MUFA and PUFA observed in the current study are in the same range as those found in two previous studies performed at 7 T.

Conclusion

The results of the current study show that it is possible to quantify the fatty acid composition of breast tissue in vivo in a clinical setting (3 T).

     

Validation of glutathione quantitation from STEAM spectra against edited 1H NMR spectroscopy at 4T: application to schizophrenia

Objective: Quantitation of glutathione (GSH) in the human brain in vivo using short echo time ¹ H NMR spectroscopy is challenging because GSH resonances are not easily resolved. The main objective of this study was to validate such quantitation in a clinically relevant population using the resolved GSH resonances provided by edited spectroscopy. A secondary objective was to compare several of the neurochemical concentrations quantified along with GSH using LCModel analysis of short echo time spectra in schizophrenia versus control. Materials and Methods: GSH was quantified at 4T from short echo STEAM spectra and MEGA-PRESS edited spectra from identical volumes of interest (anterior cingulate) in ten volunteers. Neurochemical profiles were quantified in nine controls and 13 medicated schizophrenic patients. Results: GSH concentrations as quantified using STEAM, 1.6 ± 0.4 μmol/g (mean ± SD, n=10), were within error of those quantified using edited spectra, 1.4 ± 0.4 μmol/g, and were not different (p=0.4). None of the neurochemical measurements reached sufficient statistical power to detect differences smaller than 10% in schizophrenia versus control. As such, no differences were observed. Conclusions: Human brain GSH concentrations can be quantified in a clinical setting using short-echo time STEAM spectra at 4T.     

2D and 3D texture analysis to differentiate brain metastases on MR images: proceed with caution

Objective

To find structural differences between brain metastases of lung and breast cancer, computing their heterogeneity parameters by means of both 2D and 3D texture analysis (TA).

Materials and methods

Patients with 58 brain metastases from breast (26) and lung cancer (32) were examined by MR imaging. Brain lesions were manually delineated by 2D ROIs on the slices of contrast-enhanced T1-weighted (CET1) images, and local binary patterns (LBP) maps were created from each region. Histogram-based (minimum, maximum, mean, standard deviation, and variance), and co-occurrence matrix-based (contrast, correlation, energy, entropy, and homogeneity) 2D, weighted average of the 2D slices, and true 3D TA were obtained on the CET1 images and LBP maps.

Results

For LBP maps and 2D TA contrast, correlation, energy, and homogeneity were identified as statistically different heterogeneity parameters (SDHPs) between lung and breast metastasis. The weighted 3D TA identified entropy as an additional SDHP. Only two texture indexes (TI) were significantly different with true 3D TA: entropy and energy. All these TIs discriminated between the two tumor types significantly by ROC analysis. For the CET1 images there was no SDHP at all by 3D TA.

Conclusion

Our results indicate that the used textural analysis methods may help with discriminating between brain metastases of different primary tumors.

     

In vivo 2D magnetic resonance spectroscopy of small animals

Localized in vivo NMR spectroscopy, chemical shift imaging or multi-voxel spectroscopy are potentially useful tools in small animals that are complementary to MRI, adding biochemical information to the mainly anatomical data provided by imaging of water protons. However the contribution of such methods remains hampered by the low spectral resolution of the in vivo 1D spectra. Two-dimensional methods widely developed for in vitro studies have been proposed as suitable approaches to overcome these limitations in resolution. The different homonuclear and heteronuclear sequences adapted to in vivo studies are reviewed. Their specific contributions to the spectral resolution of spectroscopic data and their limitations for in vivo investigations are discussed. The applications to experimental models of pathological processes or pharmacological treatment in mainly brain and muscle are presented. According to their combined sensitivity, acquisition duration and spatial resolution, the heteronuclear 2D experiments, which are mainly used for ¹H detected-¹³C spectroscopy after administration of ¹³C-labeled compounds, appear to be less efficient than ¹H detected-¹³C 1D methods at high field. However, the applications of 2D proton homonuclear methods show that they remain the best tools for in vivo studies when an improved resolution is required.     

Glycogen resynthesis in liver and muscle after exercise: measurement of the rate of resynthesis by13C magnetic resonance spectroscopy

We have used natural abundance¹³C magnetic resonance spectroscopy (MRS) to measure glycogen content of muscle and liver before and after heavy exercise, and after consumption of different carbohydrate-based drinks. After an overnight fast, five healthy men (mean±SEM age 23±1 years) exercised to exhaustion at 75% of VO_2max on two occasions (mean work rate 165±8 W for 78±14 min) and then, in a single blind random order, consumed either of two drinks containing the same carbohydrate load (177 g). Spectra were recorded over Vastus Lateralis muscle and the liver before and after exercise, and hourly for 5 h after the carbohydrate load. In muscle, glycogen content after exercise was 37% and 31% of basal (preexercise) concentration before consuming the drinks. After carbohydrate loading, glycogen concentration had increased significantly (p<0.05) to 70% and 64% of basal concentration respectively after 5 h. Hepatic glycogen concentration did not change significantly throughout. The study demonstrates the feasibility of sequential MRS measurement of muscle and liver glycogen before and after exercise and after carbohydrate loading.     

MRS studies of cancer

Cancer was an obvious disease to study by magnetic resonance spectroscopy (MRS); it produces large lesions that give clearly abnormal spectra, all treatment methods leave much to be desired, and radiotherapy, in particular, is limited by tissue hypoxia, a process that can be investigated by MRS.³¹P MRS has shown that tumor cells are not acidic, as had been thought; instead, the pH gradient across the tumor cell membrane is the reverse of that in a normal cell. This change in hydrogen-ion gradient is accompanied by changes in gradients of many other ions. Tumor oxygenation can be monitored in animal tumor models using the techniques employed for functional magnetic resonance imaging (MRI) of the brain. Large changes in signal are observed when drugs that reduce tumor blood flow are administered. ¹H NMR spectra of acid extracts of tumor or normal tissue biopsies contain sufficient information to permit classification (and thus, perhaps, diagnosis) if computer-based pattern recognition techniques are employed. Surprisingly, the same technique gives quite good classification of³¹P spectra takenin vivo. Can MRS beappliedin cancer therapy? Studies on tumor ion balance will help in the design of anticancer drugs and other therapies. Tumor blood flow studies using MRI could be applied to individual patients to predict the usefulness of radiotherapy and to assist in radiotherapy planning. Pattern recognition methods could automate the screening of biopsies and could also assist in interpretation of human spectra takenin vivo.     

ESMRMB 2009 Congress,Antalya, Turkey, 1–3 October: Abstracts,Thursday

Object  To determine whether glycine can be measured at 7 T in human brain with ¹H magnetic resonance spectroscopy (MRS). Materials and methods  The glycine singlet is overlapped by the larger signal of myo-inositol. Density matrix simulations were performed to determine the TE at which the myo-inositol signal was reduced the most, following a single spin-echo excitation. ¹H MRS was performed on an actively shielded 7 T scanner, in five healthy volunteers. Results  At the TE of 30 ms, the myo-inositol signal intensity was substantially reduced. Quantification using LCModel yielded a glycine-to-creatine ratio of 0.14 ± 0.01, with a Cramér–Rao lower bound (CRLB) of 7 ± 1%. Furthermore, quantification of metabolites other than glycine was possible as well, with a CRLB mostly below 10%. Conclusion  It is possible to detect glycine at 7 T in human brain, at the short TE of 30 ms with a single spin-echo excitation scheme.     

The magnitude of signal errors introduced by ISIS in quantitative31P MRS

It is well known that the quality of a quantitative³¹P MRS measurement relies largely on the performance of the volume selection method, and that image selected in vivo spectroscopy (ISIS) suffers from contaminating signal caused mostly by Tl smearing. However, these signal errors and their magnitude are seldom addressed in clinical studies. The aim of this study was therefore to investigate the magnitude of signal errors in³¹P MRS when using ISIS. The results from the measurements with a homogeneous head phantom are as follows: at low TR/T1 ratios the contamination increases rapidly, especially for small (< 27 cm³) VOI sizes; at TR/T1 = 1, the signal from a 27 cm³ VOI was 20% too high, and from an 8 cm³ VOI 150% too high. The signal obtained from different VOI positions varied between 80 and 127%. The signal varied linearly with the³¹P concentration in the object. However, a too high signal was obtained when the concentration was lower in the region of interest (inner container) than in the rest of the phantom. The agreement between the simulations and measurements shows that the results of this study are generally applicable to the measurement geometry and the ISIS experiment order rather than being specific for the MR system studied. The errors obtained both experimentally and in computer simulations are too large to be ignored in clinical studies using the ISIS pulse sequence.     

Comparison of in vivo1H MRS of human brain tumours with1H HR-MAS spectroscopy of intact biopsy samples in vitro

High resolution magic angle spinning (MAS)¹H nuclear magnetic resonance (NMR) spectroscopy has been employed to study intact human brain tumour tissue and comparison with the corresponding in vivo spectrum has been made. Two dimensional¹H MAS-NMR measurements, including J-resolved and homonuclear shift correlation spectra, were obtained to aid metabolite signal assignment. MAS gave greatly improved line-shape and reduced line-width in comparison to conventional high resolution in vivo¹H MRS of intact tissue, permitting the simultaneous detection of cellular lipids and metabolites. The technique provides the most direct method for comparison of in vivo spectra with high resolution spectra in vitro and hence allows more reliable peak assignment of in vivo¹H MRS spectra.     

Long-term sequestration of fluorinated compounds in tissues after fluvoxamine or fluoxetine treatment: a fluorine magnetic resonance spectroscopy study in vivo

Fluorine magnetic resonance spectroscopy (¹⁹F MRS), spectroscopic imaging (MRSI) and proton anatomical magnetic resonance imaging (¹H MRI) were performed on brains and lower extremities of six subjects in vivo concurrently with HPLC of serum to investigate tissue and plasma drug localization and withdrawal kinetics in humans treated with fluvoxamine or fluoxetine. ¹⁹F MRS signal was unexpectedly detected in the lower extremities months after complete disappearance of signal from plasma and brain. MRSI suggested that the lower extremity fluvoxamine signal originated mainly from bone marrow. Results suggest long-term sequestration of these drugs or their metabolites mainly in bone marrow and possibly in surrounding tissue and demonstrate the usefulness of MRS to reveal drug-trapping compartments in the body.