Understanding diffusion-weighted MRI analysis: Repeatability and performance of diffusion models in a benign breast lesion cohort.

Diffusion-weighted MRI (DWI) is an important tool for oncology research, with great clinical potential for the classification and monitoring of breast lesions. The utility of parameters derived from DWI, however, is influenced by specific analysis choices. The purpose of this study was to critically evaluate repeatability and curve-fitting performance of common DWI signal representations, for a prospective cohort of patients with benign breast lesions. Twenty informed, consented patients with confirmed benign breast lesions underwent repeated DWI (3 T) using: sagittal single-shot spin-echo echo planar imaging, bipolar encoding, TR/TE: 11,600/86 ms, FOV: 180 x 180 mm, matrix: 90 x 90, slices: 60 x 2.5 mm, iPAT: GRAPPA 2, fat suppression, and 13 b-values: 0-700 s/mm2 . A phase-reversed scan (b = 0 s/mm2 ) was acquired for distortion correction. Voxel-wise repeat-measures coefficients of variation (CoVs) were derived for monoexponential (apparent diffusion coefficient [ADC]), biexponential (intravoxel incoherent motion: f, D, D*) and stretched exponential (α, DDC) across the parameter histograms for lesion regions of interest (ROIs). Goodness-of-fit for each representation was assessed by Bayesian information criterion. The volume of interest (VOI) definition was repeatable (CoV 13.9%). Within lesions, and across both visits and the cohort, there was no dominant best-fit model, with all representations giving the best fit for a fraction of the voxels. Diffusivity measures from the signal representations (ADC, D, DDC) all showed good repeatability (CoV < 10%), whereas parameters associated with pseudodiffusion (f, D*) performed poorly (CoV > 50%). The stretching exponent α was repeatable (CoV < 12%). This pattern of repeatability was consistent over the central part of the parameter percentiles. Assumptions often made in diffusion studies about analysis choices will influence the detectability of changes, potentially obscuring useful information. No single signal representation prevails within or across lesions, or across repeated visits; parameter robustness is therefore a critical consideration. Our results suggest that stretched exponential representation is more repeatable than biexponential, with pseudodiffusion parameters unlikely to provide clinically useful biomarkers.

In vivo MR spectroscopy predicts high tumor grade in endometrial cancer.

Background In vivo magnetic resonance spectroscopy (MRS) enables non-invasive measurements of tumor metabolites. Choline-containing metabolites play a key role in tumor metabolism. Purpose To explore whether preoperative MRS-derived tumor choline levels are associated with clinical and histological features in endometrial carcinomas. Material and Methods Preoperative pelvic magnetic resonance imaging (MRI) (1.5T), including structural and diffusion-weighted imaging and localized multivoxel proton MR (1H-MR) spectroscopy, was performed in 77 prospectively included patients with histologically confirmed endometrial carcinomas. Relative levels of total choline-containing metabolites (tCho) in tumor and myometrium were measured using the ratios: tCho/Creatine; tCho/Water; and tCho/Noise. MRS parameters were analyzed in relation to histological subtype and grade, surgicopathological staging parameters, MRI-measured tumor volume, and tumor apparent diffusion coefficient (ADC) value and clinical outcome. Results Tumor tissue had significantly higher ratios for tCho/Creatine, tCho/Water, and tCho/Noise than normal myometrial tissue ( P < 0.001 for all). High tumor tCho/Water ratio was significantly associated with high tumor grade in endometrioid tumors ( P = 0.02). Tumor tCho/Creatine ratio was positively correlated to MRI-measured tumor volume (rs = 0.25; P = 0.03). Conclusion High choline levels in tumor are associated with high-risk features. In vivo MRS may potentially aid in the preoperative risk stratification in endometrial cancer.

Support vector machine for breast cancer classification using diffusion-weighted MRI histogram features: Preliminary study.

BACKGROUND: Diffusion-weighted MRI (DWI) is currently one of the fastest developing MRI-based techniques in oncology. Histogram properties from model fitting of DWI are useful features for differentiation of lesions, and classification can potentially be improved by machine learning. PURPOSE: To evaluate classification of malignant and benign tumors and breast cancer subtypes using support vector machine (SVM). STUDY TYPE: Prospective. SUBJECTS: Fifty-one patients with benign (n = 23) and malignant (n = 28) breast tumors (26 ER+, whereof six were HER2+). FIELD STRENGTH/SEQUENCE: Patients were imaged with DW-MRI (3T) using twice refocused spin-echo echo-planar imaging with echo time / repetition time (TR/TE) = 9000/86 msec, 90 × 90 matrix size, 2 × 2 mm in-plane resolution, 2.5 mm slice thickness, and 13 b-values. ASSESSMENT: Apparent diffusion coefficient (ADC), relative enhanced diffusivity (RED), and the intravoxel incoherent motion (IVIM) parameters diffusivity (D), pseudo-diffusivity (D*), and perfusion fraction (f) were calculated. The histogram properties (median, mean, standard deviation, skewness, kurtosis) were used as features in SVM (10-fold cross-validation) for differentiation of lesions and subtyping. STATISTICAL TESTS: Accuracies of the SVM classifications were calculated to find the combination of features with highest prediction accuracy. Mann-Whitney tests were performed for univariate comparisons. RESULTS: For benign versus malignant tumors, univariate analysis found 11 histogram properties to be significant differentiators. Using SVM, the highest accuracy (0.96) was achieved from a single feature (mean of RED), or from three feature combinations of IVIM or ADC. Combining features from all models gave perfect classification. No single feature predicted HER2 status of ER + tumors (univariate or SVM), although high accuracy (0.90) was achieved with SVM combining several features. Importantly, these features had to include higher-order statistics (kurtosis and skewness), indicating the importance to account for heterogeneity. DATA CONCLUSION: Our findings suggest that SVM, using features from a combination of diffusion models, improves prediction accuracy for differentiation of benign versus malignant breast tumors, and may further assist in subtyping of breast cancer. LEVEL OF EVIDENCE: 3 Technical Efficacy: Stage 3 J. Magn. Reson. Imaging 2018;47:1205-1216.

A Simplified Approach to Measure the Effect of the Microvasculature in Diffusion-weighted MR Imaging Applied to Breast Tumors: Preliminary Results.

Purpose To evaluate the relative change of the apparent diffusion coefficient (ADC) at low- and medium-b-value regimens as a surrogate marker of microcirculation, to study its correlation with dynamic contrast agent-enhanced (DCE) magnetic resonance (MR) imaging-derived parameters, and to assess its potential for differentiation between malignant and benign breast tumors. Materials and Methods Ethics approval and informed consent were obtained. From May 2013 to June 2015, 61 patients diagnosed with either malignant or benign breast tumors were prospectively recruited. All patients were scanned with a 3-T MR imager, including diffusion-weighted imaging (DWI) and DCE MR imaging. Parametric analysis of DWI and DCE MR imaging was performed, including a proposed marker, relative enhanced diffusivity (RED). Spearman correlation was calculated between DCE MR imaging and DWI parameters, and the potential of the different DWI-derived parameters for differentiation between malignant and benign breast tumors was analyzed by dividing the sample into equally sized training and test sets. Optimal cut-off values were determined with receiver operating characteristic curve analysis in the training set, which were then used to evaluate the independent test set. Results RED had a Spearman rank correlation of 0.61 with the initial area under the curve calculated from DCE MR imaging. Furthermore, RED differentiated cancers from benign tumors with an overall accuracy of 90% (27 of 30) on the test set with 88.2% (15 of 17) sensitivity and 92.3% (12 of 13) specificity. Conclusion This study presents promising results introducing a simplified approach to assess results from a DWI protocol sensitive to the intravoxel incoherent motion effect by using only three b values. This approach could potentially aid in the differentiation, characterization, and monitoring of breast pathologies. © RSNA, 2016 Online supplemental material is available for this article.

Diffusion weighted imaging for the differentiation of breast tumors: From apparent diffusion coefficient to high order diffusion tensor imaging.

BACKGROUND: To compare "standard" diffusion weighted imaging, and diffusion tensor imaging (DTI) of 2(nd) and 4(th) -order for the differentiation of malignant and benign breast lesions. METHODS: Seventy-one patients were imaged at 3 Tesla with a 16-channel breast coil. A diffusion weighted MRI sequence including b = 0 and b = 700 in 30 directions was obtained for all patients. The image data were fitted to three different diffusion models: isotropic model - apparent diffusion coefficient (ADC), 2(nd) -order tensor model (the standard model used for DTI) and a 4(th) -order tensor model, with increased degrees of freedom to describe anisotropy. The ability of the fitted parameters in the different models to differentiate between malignant and benign tumors was analyzed. RESULTS: Seventy-two breast lesions were analyzed, out of which 38 corresponded to malignant and 34 to benign tumors. ADC (using any model) presented the highest discriminative ability of malignant from benign tumors with a receiver operating characteristic area under the curve (AUC) of 0.968, and sensitivity and specificity of 94.1% and 94.7% respectively for a 1.33 × 10(-3) mm(2) /s cutoff. Anisotropy measurements presented high statistical significance between malignant and benign tumors (P < 0.001), but with lower discriminative ability of malignant from benign tumors than ADC (AUC of 0.896 and 0.897 for fractional anisotropy and generalized anisotropy respectively). Statistical significant difference was found between generalized anisotropy and fractional anisotropy for cancers (P < 0.001) but not for benign lesions (P = 0.87). CONCLUSION: While anisotropy parameters have the potential to provide additional value for breast applications as demonstrated in this study, ADC exhibited the highest differentiation power between malignant and benign breast tumors.

Metabolic profiles of brain metastases.

Metastasis to the brain is a feared complication of systemic cancer, associated with significant morbidity and poor prognosis. A better understanding of the tumor metabolism might help us meet the challenges in controlling brain metastases. The study aims to characterize the metabolic profile of brain metastases of different origin using high resolution magic angle spinning (HR-MAS) magnetic resonance spectroscopy (MRS) to correlate the metabolic profiles to clinical and pathological information. Biopsy samples of human brain metastases (n = 49) were investigated. A significant correlation between lipid signals and necrosis in brain metastases was observed (p < 0.01), irrespective of their primary origin. The principal component analysis (PCA) showed that brain metastases from malignant melanomas cluster together, while lung carcinomas were metabolically heterogeneous and overlap with other subtypes. Metastatic melanomas have higher amounts of glycerophosphocholine than other brain metastases. A significant correlation between microscopically visible lipid droplets estimated by Nile Red staining and MR visible lipid signals was observed in metastatic lung carcinomas (p = 0.01), indicating that the proton MR visible lipid signals arise from cytoplasmic lipid droplets. MRS-based metabolomic profiling is a useful tool for exploring the metabolic profiles of metastatic brain tumors.

Differentiating diffuse World Health Organization grade II and IV astrocytomas with ex vivo magnetic resonance spectroscopy.

BACKGROUND: The prognosis and treatment of astrocytomas, which are primary brain tumors, vary depending on the grade of the tumor, necessitating a precise preoperative classification. Magnetic resonance spectroscopy (MRS) provides information about metabolites in tissues and is an emerging noninvasive tool to improve diagnostic accuracy in patients with intracranial neoplasia. OBJECTIVE: To investigate whether ex vivo MRS could differentiate World Health Organization grade II (A-II) and IV astrocytomas (glioblastomas; GBM) and to correlate MR spectral profiles with clinical parameters. METHODS: Patients with A-II and GBM (n = 58) scheduled for surgical resection were enrolled. Tumor specimens were collected during surgery and stored in liquid nitrogen before being analyzed with high-resolution magic angle spinning MRS. The tumors were histopathologically classified according to World Health Organization criteria as GBM (n = 48) and A-II (n = 10). RESULTS: Multivariate analysis of ex vivo proton high-resolution magic angle spinning spectra MRS showed differences in the metabolic profiles of different grades of astrocytomas. A-II had higher levels of glycerophosphocholine and myo-inositol than GBM. The latter had more phosphocholine, glycine, and lipids. We observed a significant metabolic difference between recurrent and nonrecurrent GBM (P < .001). Primary GBM had more phosphocholine than recurrent GBM. A significant correlation (P < .001) between lipid and lactate signals and histologically estimated percentage of necrosis was observed in GBM. Spectral profiles were not correlated with age, survival, or magnetic resonance imaging-defined tumor volume. CONCLUSION: Ex vivo MRS can differentiate astrocytomas based on their metabolic profiles.

Magnetic resonance metabolomics of intact tissue: a biotechnological tool in cancer diagnostics and treatment evaluation.

Personalized medicine is increasingly important in cancer treatment for its role in staging and its potential to improve stratification of patients. Different types of molecules, genes, proteins, and metabolites are being extensively explored as potential biomarkers. This review discusses the major findings and potential of tissue metabolites determined by high-resolution magic angle spinning magnetic resonance spectroscopy for cancer detection, characterization, and treatment monitoring.

Short echo time 1H MRSI of the human brain at 3T with adiabatic slice-selective refocusing pulses; reproducibility and variance in a dual center setting.

PURPOSE: To assess the reproducibility of (1)H-MR spectroscopic imaging (MRSI) of the human brain at 3T with volume selection by a double spin echo sequence for localization with adiabatic refocusing pulses (semi-LASER). MATERIALS AND METHODS: Twenty volunteers in two different institutions were measured twice with the same pulse sequence at an echo time of 30 msec. Magnetic resonance (MR) spectra were analyzed with LCModel with a simulated basis set including an experimentally acquired macromolecular signal profile. For specific regions in the brain mean metabolite levels, within and between subject variance, and the coefficient of variation (CoV) were calculated (for taurine, glutamate, total N-acetylaspartate, total creatine, total choline, myo-inositol + glycine, and glutamate + glutamine). RESULTS: Repeated measurements showed no significant differences with a paired t-test and a high reproducibility (CoV ranging from 3%-30% throughout the selected volume). Mean metabolite levels and CoV obtained in similar regions in the brain did not differ significantly between two contributing institutions. The major source of differences between different measurements was identified to be the between-subject variations in the volunteers. CONCLUSION: We conclude that semi-LASER (1)H-MRSI at 3T is an adequate method to obtain quantitative and reproducible measures of metabolite levels over large parts of the brain, applicable across multiple centers.

Characterization of brain metastases using high-resolution magic angle spinning MRS.

The objectives of this study were to (a) explore the spectral characteristics of brain metastases, focusing on the origin of the primary cancer, and (b) evaluate the correlation with clinical outcome using multivariate analysis. High-resolution magic angle spinning (HR-MAS) MR spectra (n = 26) were obtained from 16 patients with brain metastases using a Bruker Avance DRX600 instrument. Standard pulse-acquired and spin-echo (TE 32 and 285 ms) (1)H spectra were obtained. These were examined using principal component analysis (PCA) and partial least squares regression analysis (PLS) relating spectral data to clinical outcome. The PCA score plot of pulse-acquired HR-MAS spectra showed a trend of clustering due to the origin of the metastases, mainly based on differences in the lipid signals at 1.3 and 0.9 ppm. With PLS, spectra of patients who died less than 5 months after surgery appeared to cluster in the lower left quadrant of the score plot. These preliminary results on brain metastasis classification and prediction of survival must be validated in a larger patient cohort. However, the possibility of differentiating metastases according to origin and predicting survival on the basis of HR-MAS spectra suggests that this method may be useful for diagnosing and planning treatment for brain metastases and also for guiding decisions about terminating further treatment.

Metabolic profiling of human brain metastases using in vivo proton MR spectroscopy at 3T.

BACKGROUND: Metastases to the central nervous system from different primary cancers are an oncologic challenge as the overall prognosis for these patients is generally poor. The incidence of brain metastases varies with type of primary cancer and is probably increasing due to improved therapies of extracranial metastases prolonging patient's overall survival and thereby time for brain metastases to develop. In addition, the greater access to improved neuroimaging techniques can provide earlier diagnosis. The aim of this study was to investigate the feasibility of using proton magnetic resonance spectroscopy (MRS) and multivariate analyses to characterize brain metastases originating from different primary cancers, to assess changes in spectra during radiation treatment and to correlate the spectra to clinical outcome after treatment. METHODS: Patients (n = 26) with brain metastases were examined using single voxel MRS at a 3T clinical MR system. Five patients were excluded due to poor spectral quality. The spectra were obtained before start (n = 21 patients), immediately after (n = 6 patients) and two months after end of treatment (n = 4 patients). Principal component analysis (PCA) and partial least square regression analysis (PLS) were applied in order to identify clustering of spectra due to origin of metastases and to relate clinical outcome (survival) of the patients to spectral data from the first MR examination. RESULTS: The PCA results indicated that brain metastases from primary lung and breast cancer were separated into two clusters, while the metastases from malignant melanomas showed no uniformity. The PLS analysis showed a significant correlation between MR spectral data and survival five months after MRS before start of treatment. CONCLUSION: MRS determined metabolic profiles analysed by PCA and PLS might give valuable clinical information when planning and evaluating the treatment of brain metastases, and also when deciding to terminate further therapies.

Cerebral metabolite differences in adolescents with low birth weight: assessment with in vivo proton MR spectroscopy.

BACKGROUND: Children with very low birth weight (VLBW) have a significantly increased risk of later neurodevelopmental problems, while infants born small for gestational age (SGA) at term are also at some risk of developing neurological impairment. OBJECTIVE: To investigate possible brain metabolite differences in adolescents with VLBW, SGA at term and controls by proton in vivo magnetic resonance spectroscopy (MRS) at 1.5 T. MATERIALS AND METHODS: MR spectra were acquired from volumes localized in the left frontal lobe, containing mainly white matter (54 subjects). Peak areas of N-acetyl aspartate (NAA), choline (Cho) and creatine (Cr) were determined, and the peak area ratio of NAA to Cr, total Cho to Cr, or NAA to Cho calculated. Probabilistic neural network (PNN) analysis was performed utilizing the chemical shift region containing resonances from NAA, Cho and Cr as inputs. RESULTS: No significant difference in the peak area ratios could be found using the Kruskal-Wallis test. By application of PNN, a correct classification of 52 of the 54 adolescents with a sensitivity and specificity exceeding 93% for all groups was achieved. CONCLUSION: Small, yet systematic, differences in brain metabolite distribution among the groups were confirmed by PNN analysis.

Manganese, copper, and zinc in cerebrospinal fluid from patients with multiple sclerosis.

The concentrations of manganese, copper, and zinc in cerebrospinal fluid (CSF) from patients with multiple sclerosis (MS) and patients with no known neurological disease (control group) were measured. Manganese and copper levels were determined by two different analytical methods: atomic absorption spectrometry (AAS) and high-resolution inductively coupled plasma-mass spectrometry (HR-ICP-MS), whereas zinc levels were determined by HR-ICP-MS only. Manganese levels (mean+/-SEM) were significantly decreased in the CSF of MS patients (1.07+/-0.13 microg/L, ICP-MS; 1.08+/-0.11 microg/L, AAS) compared to the levels in the control group (1.78+/-0.26 microg/L, ICP-MS; 1.51+/-0.17 microg/L, AAS). Copper levels were significantly elevated in the CSF of MS patients (10.90+/-1.11 microg/L; ICP-MS, 11.53+/-0.83 microg/L, AAS) compared to the levels in the control group (8.67+/-0.49 microg/L, ICP-MS; 9.10+/-0.62 microg/L, AAS). There were no significant differences between the CSF zinc levels of MS and control patients. The physiological basis for the differences in manganese and copper concentrations between MS patients and controls is unknown, but could be related to alterations in the manganese- containing enzyme glutamine synthetase and the copper-containing enzyme cytochrome oxidase.

Concentrations of Cd, Co, Cu, Fe, Mn, Rb, V, and Zn in formalin-fixed brain tissue in amyotrophic lateral sclerosis and Parkinsonism-dementia complex of Guam determined by High-resolution ICP-MS.

Amyotrophic lateral sclerosis (ALS) and parkinsonism-dementia complex (PDC) are neurodegenerative disorders that occurred with extremely high frequency among the native population on Guam, especially in the 1950s and 1960s, but have substantially declined over the last half-century. The etiology of these diseases is unknown, but the most plausible hypothesis centers on imbalances in essential and toxic metals. We have determined the concentrations of Cd, Co, Cu, Fe, Mn, Rb, V, and Zn in formalin-fixed brain tissue collected during the period 1979-1983 from eight Guamanian patients with ALS, four with PDC, and five control subjects using high-resolution inductively coupled plasma-mass spectrometry. The concentrations of Cd are markedly and significantly elevated both in gray and white matter in ALS, but not in PDC patients. The concentrations of Zn are elevated for both patient groups, in both gray and white matter, but only the difference in gray matter for PDC is significant. For the other metals, no significant differences are found.