Dynamic oxygen challenge evaluated by NMR T1 and T2* – insights into tumor oxygenation

There is intense interest in developing non‐invasive prognostic biomarkers of tumor response to therapy, particularly with regard to hypoxia. It has been suggested that oxygen sensitive MRI, notably blood oxygen level‐dependent (BOLD) and tissue oxygen level‐dependent (TOLD) contrast, may provide relevant measurements. This study examined the feasibility of interleaved T₂*‐ and T₁‐weighted oxygen sensitive MRI, as well as R₂* and R₁ maps, of rat tumors to assess the relative sensitivity to changes in oxygenation. Investigations used cohorts of Dunning prostate R3327‐AT1 and R3327‐HI tumors, which are reported to exhibit distinct size‐dependent levels of hypoxia and response to hyperoxic gas breathing. Proton MRI R₁ and R₂* maps were obtained for tumors of anesthetized rats (isoflurane/air) at 4.7 T. Then, interleaved gradient echo T₂*‐ and T₁‐weighted images were acquired during air breathing and a 10 min challenge with carbogen (95% O₂–5% CO₂). Signals were stable during air breathing, and each type of tumor showed a distinct signal response to carbogen. T₂* (BOLD) response preceded T₁ (TOLD) responses, as expected. Smaller HI tumors (reported to be well oxygenated) showed the largest BOLD and TOLD responses. Larger AT1 tumors (reported to be hypoxic and resist modulation by gas breathing) showed the smallest response. There was a strong correlation between BOLD and TOLD signal responses, but ΔR₂^* and ΔR₁ were only correlated for the HI tumors. The magnitude of BOLD and TOLD signal responses to carbogen breathing reflected expected hypoxic fractions and oxygen dynamics, suggesting potential value of this test as a prognostic biomarker of tumor hypoxia. Copyright © 2015 John Wiley & Sons, Ltd.     

A simple breathing circuit to maintain isocapnia during measurements of the hypoxic ventilatory response

We report the development and testing of a simple breathing circuit that maintains isocapnia in human subjects during hypoxic hyperpnea. In addition, the circuit permits rapid switching between two gas mixtures with different partial pressures of oxygen. Eleven volunteers breathed repeated cycles of exposure to air (2 min of 21% O(2), balance N(2)) and hypoxia (2 min of 8.3+/-0.1% O(2), balance N(2)). Hypoxia induced significant increases in minute ventilation, breathing frequency and tidal volume (P < 0.05) that were consistent over repeated cycles of hypoxia (P > 0.1, one-way ANOVA). The system successfully maintained isocapnia in all subjects, with an average change in end-tidal CO(2) of only -0.2 mmHg during hyperventilation in hypoxia (range 0.4 to -0.8 mmHg). This system may be suitable for repeated tests of the hypoxic ventilatory response (HVR) and may prove useful for exploring intra- and inter-individual variability of HVR in humans.     

Oxygen‐induced frequency shifts in hyperoxia: a significant component of BOLD signal

In comparison to the well‐documented significance of intravascular deoxyhemoglobin (deoxyHgb), the effects of dissolved oxygen on the blood‐oxygen‐level‐dependent (BOLD) signal have not been widely reported. Based on the fact that the prolonged inspiration of high oxygen fraction gas can result in up to a sixfold increase of the baseline tissue oxygenation, the current study focused on the influence of dissolved oxygen on the BOLD signal during hyperoxia. As results, our in vitro study revealed that the r₁ and r₂ (relaxivities) of the oxygen‐treated serum were 0.22 mM^?1 · s^?1and 0.19 mM^?1 · s^?1_, respectively. In an in vivo experiment, hyperoxic respiration induced negative BOLD contrast (i.e. signal decrease) in 18–42% of measured brain regions, voxels with accompanying significant decreases in both the T₂^* (?12.1% to ?19.4%) and T₁ (?5.8% to ?3.3%) relaxation times. In contrast, the T₂^* relaxation time significantly increased (11.2% to 14.0%) for the voxels displaying positive BOLD contrast (in 41–50% of the measured brain), which reflected a hyperoxygenation‐induced reduction in tissue deoxyHgb concentration. These data imply that hyperoxia‐driven BOLD signal changes are primarily determined by the counteracting effects of extravascular oxygen and intravascular deoxyHgb. Oxygen‐induced magnetic susceptibility was further demonstrated by the study of 1 min hypoxia, which induced BOLD signal changes opposite to those under hyperoxia. Vasoconstriction was more common in voxels with negative BOLD contrast than in voxels with positive contrast (% change of blood volume, ?9.8% to ?12.8% versus 2.0% to 2.2%), which further suggests that negative BOLD contrast is mainly evoked by an increase in extravascular oxygen concentration. Conclusively, frequency shifts, which are induced by the accumulation of oxygen molecules and associated magnetic field inhomogeneity, are a significant source of the negative BOLD contrast during hyperoxia. Copyright © 2014 John Wiley & Sons, Ltd.     

Analysis of parametric histogram from dynamic contrast‐enhanced MRI: application in evaluating brain tumor response to radiotherapy

Dynamic contrast‐enhanced MRI (DCE MRI) has been used to study tumor response to treatment for many years. In this study, the modified full width at half‐maximum (mFWHM), calculated from the wash‐in slope histogram, is proposed as a parameter for the evaluation of changes in tumor heterogeneity which respond to radiotherapy. Twenty‐five patients with brain tumors were evaluated and divided into the nonresponder group (n = 11) and the responder group (n = 14) according to the Response Evaluation Criteria in Solid Tumors (RECIST). All selected tumors were evaluated by mFWHM ratios of post‐ to pre‐therapy (the ratio was defined as the therapeutic mFWHM ratio, TMR). The changes in kurtosis of the histograms and the averaged K^trans within a tumor were also calculated for comparison. The receiver operating characteristic analysis and Kaplan–Meier curves were used to examine the diagnosis ability. The TMR values were significantly higher in nonresponders than in responders (p < 0.001). When compared with the other two parameters, the proposed method also demonstrated better sensitivity and specificity. When adopting the TMR for the estimation of prognosis after therapy, there was a significant difference between the population survival curves. In conclusion, the derived mFWHM reflects tumor heterogeneity, and the ability to depict patient survival probability from TMR corresponds well with that from RECIST. The results reveal that, in brain tumors, progression may be exhibited not only by tumor size, but also by tumor heterogeneity. Copyright © 2012 John Wiley & Sons, Ltd.     

Effect of anesthesia on renal R2* measured by blood oxygen level‐dependent MRI

Blood oxygen level‐dependent (BOLD) MRI is increasingly being used to assess renal tissue oxygenation during disease based on the transverse relaxation rate (R₂*). In preclinical small animal models, the requisite use of anesthesia during imaging may lead to functional changes which influence R₂* and confound results. The purpose of this study was to evaluate the effects of four common anesthetic compounds on renal R₂* in healthy mice. Five female ICR mice were imaged with BOLD MRI approximately 25 min after induction with isoflurane (Iso; 1% or 1.5%, delivered in 100% O₂), ketamine/xylazine (KX), sodium pentobarbital (PB) or 2,2,2‐tribromoethanol (TBE). A significant effect of anesthetic agent on R₂* was observed in all tissue layers of the kidney, including the cortex, outer stripe of the outer medulla (OSOM), inner stripe of the outer medulla (ISOM) and inner medulla (IM). Pairwise significant differences in R₂* between specific agents were found in the cortex, OSOM and ISOM, with the largest difference observed in the ISOM between 1.5% Iso (26.6 ± 1.7 s^–1) and KX (66.0 ± 7.1 s^–1). The difference between 1% Iso and KX in the ISOM was not abolished when KX was administered with supplemental 100% O₂ or when 1% Iso was delivered in 21% O₂, indicating that the fraction of inspired oxygen did not account for the observed differences. Our results indicate that the choice of anesthesia has a large influence on the observed R₂* in mouse kidney, and anesthetic effects must be considered in the design and interpretation of renal BOLD MRI studies. Copyright © 2015 John Wiley & Sons, Ltd.     

Quantitative blood oxygenation level‐dependent (BOLD) response of the left ventricular myocardium to hyperoxic respiratory challenge at 1.5 and 3.0 T

The aim of this study was to quantify the response of the myocardial transverse relaxation times (ΔT₂*) to hyperoxic respiratory challenge (HRC) at different field strengths in an intra‐individual comparison of healthy volunteers and in a patient with coronary artery disease. Blood oxygenation level‐dependent (BOLD) cardiovascular MR (CMR) data were acquired in 10 healthy volunteers (five women, five men; mean age, 29 ± 3 years; range, 22–35 years) at 1.5 and 3.0 T. Medical air (21% O₂), pure oxygen and carbogen (95% O₂, 5% CO₂) were administered in a block‐design temporal pattern to induce normoxia, hyperoxia and hyperoxic hypercapnia, respectively. Average T₂* times were derived from measurements by two independent and blind readers in 16 standard myocardial segments on three short‐axis slices per patient. Inter‐ and intra‐reader correlations of T₂* measurements were good [intra‐class correlation coefficient (ICC) = 0.75 and ICC = 0.79, both p < 0.001]. During normoxia, the mean T₂* times were 29.9 ± 6.1 ms at 1.5 T and 27.1 ± 6.6 ms at 3.0 T. Both hyperoxic gases induced significant (all p < 0.01) T₂* increases (?T₂* hyperoxia: 1.5 T, 12.7%; 3.0 T, 11.2%; hyperoxic hypercapnia: 1.5 T, 13.1%; 3.0 T, 17.7%). Analysis of variance (ANOVA) results indicated a significant (both p < 0.001) effect of the inhaled gases on the T₂* times at both 1.5 T (F = 17.74) and 3.0 T (F = 39.99). With regard to the patient imaged at 1.5 T, HRC induced significant T₂* increases during hyperoxia and hyperoxic hypercapnia in normal myocardial segments, whereas the T₂* response was not significant in ischemic segments (p > 0.23). The myocardial ?T₂* response to HRC can reliably be imaged and quantified with BOLD CMR at both 1.5 and 3.0 T. During HRC, hyperoxia and hyperoxic hypercapnia induce a significant increase in T₂*, with ?T₂* being largest at 3.0 T and during hyperoxic hypercapnia in normal myocardial segments. Copyright © 2014 John Wiley & Sons, Ltd.     

Implementation of spin‐echo blood oxygen level‐dependent (BOLD) functional MRI in birds

The advent of high‐field MRI systems has allowed the implementation of blood oxygen level‐dependent functional MRI (BOLD fMRI) on small animals. An increased magnetic field improves the signal‐to‐noise ratio and thus allows an improvement in the spatial resolution. However, it also increases susceptibility artefacts in the commonly acquired gradient‐echo images. This problem is particularly prominent in songbird MRI because of the presence of numerous air cavities in the skull of birds. These T₂*‐related image artefacts can be circumvented using spin‐echo BOLD fMRI. In this article, we describe the implementation of spin‐echo BOLD fMRI in zebra finches, a small songbird of 15–25 g, extensively studied in the behavioural neurosciences of birdsong. Because the main topics in this research domain are song perception and song learning, the protocol implemented used auditory stimuli. Despite the auditory nature of the stimuli and the weak contrast‐to‐noise ratio of spin‐echo BOLD fMRI compared with gradient‐echo BOLD fMRI, we succeeded in detecting statistically significant differences in BOLD responses triggered by different stimuli. This study shows that spin‐echo BOLD fMRI is a viable approach for the investigation of auditory processing in the whole brain of small songbirds. It can also be applied to study auditory processing in other small animals, as well as other sensory modalities. Copyright © 2010 John Wiley & Sons, Ltd.     

Pushing the limits of high‐resolution functional MRI using a simple high‐density multi‐element coil design

Recent studies have shown that functional MRI (fMRI) can be sensitive to the laminar and columnar organization of the cortex based on differences in the spatial and temporal characteristics of the blood oxygenation level‐dependent (BOLD) signal originating from the macrovasculature and the neuronal‐specific microvasculature. Human fMRI studies at this scale of the cortical architecture, however, are very rare because the high spatial/temporal resolution required to explore these properties of the BOLD signal are limited by the signal‐to‐noise ratio. Here, we show that it is possible to detect BOLD signal changes at an isotropic spatial resolution as high as 0.55 mm at 7 T using a high‐density multi‐element surface coil with minimal electronics, which allows close proximity to the head. The coil comprises of very small, 1 × 2‐cm², elements arranged in four flexible modules of four elements each (16‐channel) that can be positioned within 1 mm from the head. As a result of this proximity, tissue losses were five‐fold greater than coil losses and sufficient to exclude preamplifier decoupling. When compared with a standard 16‐channel head coil, the BOLD sensitivity was approximately 2.2‐fold higher for a high spatial/temporal resolution (1 mm isotropic/0.4 s), multi‐slice, echo planar acquisition, and approximately three‐ and six‐fold higher for three‐dimensional echo planar images acquired with isotropic resolutions of 0.7 and 0.55 mm, respectively. Improvements in parallel imaging performance (geometry factor) were up to around 1.5‐fold with increasing acceleration factor, and improvements in fMRI detectability (temporal signal‐to‐noise ratio) were up to around four‐fold depending on the distance to the coil. Although deeper lying structures may not benefit from the design, most fMRI questions pertain to the neocortex which lies within approximately 4 cm from the surface. These results suggest that the resolution of fMRI (at 7 T) can approximate levels that are closer to the spatial/temporal scale of the fundamental functional organization of the human cortex using a simple high‐density coil design for high sensitivity. Copyright © 2012 John Wiley & Sons, Ltd.     

Proton and sodium MRI assessment of emerging tumor chemotherapeutic resistance

The ultimate goal of any cancer therapy is to target the elimination of neoplastic cells. Although newer therapeutic strategies are in constant development, therapeutic assessment has been hampered by the inability to assess, rapidly and quantitatively, efficacy in vivo. Diffusion imaging and, more recently, sodium MRI have demonstrated their distinct abilities to detect therapy-induced alterations in tumor cellularity, which has been demonstrated to be indicative of therapeutic efficacy. More importantly, both imaging modalities detect tumor response much earlier than traditional methodologies that rely on macroscopic volumetric changes. In this study, the correlation between tumor sodium and diffusion was further tested to demonstrate the sensitivity of sodium imaging to gauge tumor response to therapy by using a 9L rat gliosarcoma treated with varying doses of BCNU [1,3-bis(2-chloroethyl)-1-nitrosourea]. This orthotopic model has been demonstrated to display variability in response to BCNU therapy where initial insult has been shown to lead to drug-resistance. In brief, a single 26.6 mg/kg BCNU dose yielded dramatic responses in both diffusion and sodium MRI. However, a second equivalent BCNU dose yielded a much smaller change in diffusion and sodium, suggesting a drop in tumor sensitivity to BCNU. The MRI responses of animals treated with 13.3 mg/kg BCNU were much lower and similar responses were observed after the initial and secondary applications of BCNU. Furthermore, these results were further validated using volumetric measurements of the tumor and also ex vivo determination of tumor sensitivity to BCNU. Overall, these experiments demonstrate the sensitivity and applicability of sodium and diffusion MRI as tools for dynamic assessment of tumor response to therapy.     

MRI assessment of cerebral oxygen metabolism in cocaine‐addicted individuals: hypoactivity and dose dependence

Long‐term cocaine use is known to negatively impact neural and cerebrovascular systems. However, the use of imaging markers to separately assess these parameters remains challenging. The primary reason is that most functional imaging markers, such as cerebral blood flow, functional connectivity, and task‐evoked functional MRI, are known to reflect a complex interplay between neural and vascular components, thus the interpretation of the results is not straightforward. The goal of the present study is to examine neural‐activity‐specific changes in cocaine addiction, using cerebral metabolic rate of oxygen (CMRO2) as a surrogate marker of aggregated neural activity. We applied a recently developed CMRO2 technique in 13 cocaine‐addicted subjects and 13 age‐ and gender‐matched control subjects, and examined the impact of long‐term cocaine use on CMRO2. Our results showed that CMRO2 in cocaine‐addicted subjects (152 ± 16 µmol/100 g/min) is significantly lower (p = 0.031) than that in controls (169 ± 20 µmol/100 g/min). Furthermore, the severity of this decreased metabolism is associated with lifetime cocaine use (p = 0.05). Additionally, the CMRO2 reduction was accompanied by a trend of decrease in cerebral blood flow (p = 0.058), but venous oxygenation was unaffected (p = 0.96), which suggested that the CMRO2 change may be attributed to a vascular deficiency in chronic cocaine users. To our knowledge, this is the first study to measure CMRO2 in cocaine‐addicted individuals. Our findings suggest that CMRO2 may be a promising approach for assessing the long‐term effects of cocaine use on the brain. Copyright © 2014 John Wiley & Sons, Ltd.     

Assessment of tumor necrotic fraction by dynamic contrast‐enhanced MRI: a preclinical study of human tumor xenografts with histopathologic correlation

Contrary to the common notion that tumor necrotic regions are non‐enhancing after contrast administration, recent evidence has shown that necrotic regions exhibit delayed and slow uptake of gadolinium tracer on dynamic contrast‐enhanced MRI (DCE MRI). The purpose of this study is to explore whether the mapping of tumor voxels with delayed and slow enhancement on DCE MRI can be used to derive estimates of tumor necrotic fraction. Patient‐derived tumor xenograft lines of seven human cancers were implanted in 26 mice which were subjected to DCE MRI performed using a spoiled gradient recalled sequence. Gadolinium tracer concentration was estimated using the variable flip angle technique. To identify tumor voxels exhibiting delayed and slow uptake of contrast medium, clustering analysis was performed using a k‐means clustering algorithm that classified tumor voxels according to their contrast enhancement patterns. Comparison of the percentage of tumor voxels exhibiting delayed and slow enhancement with the tumor necrotic fraction estimated on histology showed a strong correlation (r = 0.962, p < 0.001). The mapping of tumor regions with delayed and slow contrast uptake on DCE MRI correlated strongly with tumor necrotic fraction, and can potentially serve as a non‐invasive imaging surrogate for the in vivo assessment of necrotic fraction. Copyright © 2014 John Wiley & Sons, Ltd.     

Characterization of a rat orthotopic pancreatic head tumor model using three‐dimensional and quantitative multi‐parametric MRI

The purpose of this study was to investigate the reliability of 3D isotropic MRI and quantitative multi‐parametric MRI characterization on an orthotopic pancreatic head tumor model in rats. 3D isotropic T₂‐weighted MRI was performed as a routine for tumor longitudinal follow‐up and volume estimation. Common bile duct diameter was measured from 3D multiplanar reconstruction. Quantitative multi‐parametric measurements including pixel‐wise T₂, T₁ relaxivity, apparent diffusion coefficient (ADC) and apparent diffusion kurtosis mapping were performed twice throughout tumor growth. Semi‐quantitative and quantitative analyses based on an extended Tofts model were applied to region‐of‐interest‐based dynamic contrast‐enhanced imaging, followed by contrast ratio measurement on standard contrast‐enhanced imaging. Moreover, low‐level texture‐based analysis was inspected for T₂, T₁, ADC and contrast ratio measurements. Results indicated that multi‐parametric MRI showed good reproducibility for tumor characterization; the measurements were not affected by tumor growth. Tumor growth was further confirmed with histology examinations. To conclude, state‐of‐the‐art clinical MRI techniques were translated to this preclinical tumor model with high reliability, and have paved the way for translational oncology studies on this tumor model.     

Longitudinal study of the assessment by MRI and diffusion-weighted imaging of tumor response in patients with locally advanced breast cancer undergoing neoadjuvant chemotherapy

Measurements of tumor apparent diffusion coefficient (ADC), volume and diameter in assessing the response of patients with locally advanced breast cancer (LABC) (n = 56) undergoing neoadjuvant chemotherapy (NACT) at four time periods (before treatment and after three cycles of NACT) were carried out at 1.5 T using diffusion-weighted imaging (DWI) and MRI. Ten benign tumors and 15 controls were also investigated. The MR tumor response was compared with the clinical response. Mean ADC before treatment of malignant breast tissue was significantly lower than that of controls, disease-free contralateral tissue of the patients, and benign lesions, and gradually increased during the course of NACT. Analysis of the percentage change in ADC, volume and diameter after each cycle of NACT between clinical responders and non-responders showed that the change in ADC after the first cycle was statistically significant compared with volume and diameter, indicating its potential in assessing early response. After the third cycle, the sensitivity for differentiating responders from non-responders was 89% for volume and diameter and 68% for ADC, and the respective specificities were 50%, 70% and 100%. A sensitivity of 84% (specificity of 60% with an accuracy of 76%) was achieved when all three variables were taken together to predict the response. A cut-off value of ADC was also calculated using receiver operator characteristics analysis to discriminate between normal, benign and malignant breast tissue. Similarly, a cut-off value for ADC, volume and diameter was obtained after the second and third cycles of NACT to predict tumor response. The results show that ADC is more useful for predicting early tumor response to NACT than morphological variables, suggesting its potential in effective treatment management.     

Diffusion‐weighted MRI monitoring of pancreatic cancer response to radiofrequency heat‐enhanced intratumor chemotherapy

The aim of this study was to evaluate the feasibility of using diffusion‐weighted MRI to monitor the early response of pancreatic cancers to radiofrequency heat (RFH)‐enhanced chemotherapy. Human pancreatic carcinoma cells (PANC‐1) in different groups and 24 mice with pancreatic cancer xenografts in four groups were treated with phosphate‐buffered saline (PBS) as a control, RFH at 42 °C, gemcitabine and gemcitabine plus RFH at 42 °C. One day before and 1, 7 and 14 days after treatment, diffusion‐weighted MRI and T₂‐weighted imaging were applied to monitor the apparent diffusion coefficients (ADCs) of tumors and tumor growth. MRI findings were correlated with the results of tumor apoptosis analysis. In the in vitro experiments, the quantitative viability assay showed lower relative cell viabilities for treatment with gemcitabine plus RFH at 42 °C relative to treatment with RFH only and gemcitabine only (37 ± 5% versus 65 ± 4% and 58 ± 8%, respectively, p < 0.05). In the in vivo experiments, the combination therapy resulted in smaller relative tumor volumes than RFH only and chemotherapy only (0.82 ± 0.17 versus 2.23 ± 0.90 and 1.64 ± 0.44, respectively, p = 0.003). In vivo, 14‐T MRI demonstrated a remarkable decrease in ADCs at day 1 and increased ADCs at days 7 and 14 in the combination therapy group. The apoptosis index in the combination therapy group was significantly higher than those in the chemotherapy‐only, RFH‐only and PBS treatment groups (37 ± 6% versus 20 ± 5%, 8 ± 2% and 3 ± 1%, respectively, p < 0.05). This study confirms that it is feasible to use MRI to monitor RFH‐enhanced chemotherapy in pancreatic cancers, which may present new options for the efficient treatment of pancreatic malignancies using MRI/RFH‐integrated local chemotherapy. Copyright © 2013 John Wiley & Sons, Ltd.     

MRI & MRS assessment of the role of the tumour microenvironment in response to therapy

MRI and MRS techniques are being applied to the characterisation of various aspects of the tumour microenvironment and to the assessment of tumour response to therapy. For example, kinetic parameters describing tumour blood vessel flow and permeability can be derived from dynamic contrast-enhanced MRI data and have been correlated with a positive tumour response to antivascular therapies. The ongoing development and validation of noninvasive, high-resolution anatomical/molecular MR techniques will equip us with the means to detect specific tumour biomarkers early on, and then to monitor the efficacy of cancer treatments efficiently and reliably, all within a clinically relevant time frame. Reliable tumour microenvironment imaging biomarkers will provide obvious advantages by enabling tumour-specific treatment tailoring and potentially improving patient outcome. However, for routine clinical application across many disease types, such imaging biomarkers must be quantitative, robust, reproducible, sufficiently sensitive and cost-effective. These characteristics are all difficult to achieve in practice, but image biomarker development and validation have been greatly facilitated by an increasing number of pertinent preclinical in vivo cancer models. Emphasis must now be placed on discovering whether the preclinical results translate into an improvement in patient care and, therefore, overall survival.     

Gaussian mixture model‐based classification of dynamic contrast enhanced MRI data for identifying diverse tumor microenvironments: preliminary results

Tumor hypoxia develops heterogeneously, affects radiation sensitivity and the development of metastases. Prognostic information derived from the in vivo characterization of the spatial distribution of hypoxic areas in solid tumors can be of value for radiation therapy planning and for monitoring the early treatment response. Tumor hypoxia is caused by an imbalance between the supply and consumption of oxygen. The tumor oxygen supply is inherently linked to its vasculature and perfusion which can be evaluated by dynamic contrast enhanced (DCE‐) MRI using the contrast agent Gd‐DTPA. Thus, we hypothesize that DCE‐MRI data may provide surrogate information regarding tumor hypoxia. In this study, DCE‐MRI data from a rat prostate tumor model were analysed with a Gaussian mixture model (GMM)‐based classification to identify perfused, hypoxic and necrotic areas for a total of ten tumor slices from six rats, of which one slice was used as training data for GMM classifications. The results of pattern recognition analyzes were validated by comparison to corresponding Ak_ep maps defining the perfused area (0.84 ± 0.09 overlap), hematoxylin and eosin (H&E)‐stained tissue sections defining necrosis (0.64 ± 0.15 overlap) and pimonidazole‐stained sections defining hypoxia (0.72 ± 0.17 overlap), respectively. Our preliminary data indicate the feasibility of a GMM‐based classification to identify tumor hypoxia, necrosis and perfusion/permeability from non‐invasively acquired, in vivo DCE‐MRI data alone, possibly obviating the need for invasive procedures, such as biopsies, or exposure to radioactivity, such as positron emission tomography (PET) exams. Copyright © 2013 John Wiley & Sons, Ltd.