Toward More Accessible Fully Automated 3D Volumetric MRI Decision Trees for the Differential Diagnosis of Multiple System Atrophy, Related Disorders, and Age-Matched Healthy Subjects.

Differentiating multiple system atrophy (MSA) from related neurodegenerative movement disorders (NMD) is challenging. MRI is widely available and automated decision-tree analysis is simple, transparent, and resistant to overfitting. Using a retrospective cohort of heterogeneous clinical MRIs broadly sourced from a tertiary hospital system, we aimed to develop readily translatable and fully automated volumetric diagnostic decision-trees to facilitate early and accurate differential diagnosis of NMDs. 3DT1 MRI from 171 NMD patients (72 MSA, 49 PSP, 50 PD) and 171 matched healthy subjects were automatically segmented using Freesurfer6.0 with brainstem module. Decision trees employing substructure volumes and a novel volumetric pons-to-midbrain ratio (3D-PMR) were produced and tenfold cross-validation performed. The optimal tree separating NMD from healthy subjects selected cerebellar white matter, thalamus, putamen, striatum, and midbrain volumes as nodes. Its sensitivity was 84%, specificity 94%, accuracy 84%, and kappa 0.69 in cross-validation. The optimal tree restricted to NMD patients selected 3D-PMR, thalamus, superior cerebellar peduncle (SCP), midbrain, pons, and putamen as nodes. It yielded sensitivities/specificities of 94/84% for MSA, 72/96% for PSP, and 73/92% PD, with 79% accuracy and 0.62 kappa. There was correct classification of 16/17 MSA, 5/8 PSP, 6/8 PD autopsy-confirmed patients, and 6/8 MRIs that preceded motor symptom onset. Fully automated decision trees utilizing volumetric MRI data distinguished NMD patients from healthy subjects and MSA from other NMDs with promising accuracy, including autopsy-confirmed and pre-symptomatic subsets. Our open-source methodology is well-suited for widespread clinical translation. Assessment in even more heterogeneous retrospective and prospective cohorts is indicated.

Neuroimaging of COVID-19.

We review the wide variety of common neuroimaging manifestations related to coronavirus disease 2019 (COVID-19) and COVID therapies, grouping the entities by likely pathophysiology, recognizing that the etiology of many entities remains uncertain. Direct viral invasion likely contributes to olfactory bulb abnormalities. COVID meningoencephalitis may represent direct viral infection and/or autoimmune inflammation. Para-infectious inflammation and inflammatory demyelination at the time of infection are likely primary contributors to acute necrotizing encephalopathy, cytotoxic lesion of the corpus callosum, and diffuse white matter abnormality. Later postinfectious inflammation and demyelination may manifest as acute demyelinating encephalomyelitis, Guillain-Barré syndrome, or transverse myelitis. The hallmark vascular inflammation and coagulopathy of COVID-19 may produce acute ischemic infarction, microinfarction contributing to white matter abnormality, space-occupying hemorrhage or microhemorrhage, venous thrombosis, and posterior reversible encephalopathy syndrome. Adverse effects of therapies including zinc, chloroquine/hydroxychloroquine, antivirals, and vaccines, and current evidence regarding "long COVID" is briefly reviewed. Finally, we present a case of bacterial and fungal superinfection related to immune dysregulation from COVID.

Clinical Trial-Ready Patient Cohorts for Multiple System Atrophy: Coupling Biospecimen and iPSC Banking to Longitudinal Deep-Phenotyping.

Multiple system atrophy (MSA) is a fatal neurodegenerative disease of unknown etiology characterized by widespread aggregation of the protein alpha-synuclein in neurons and glia. Its orphan status, biological relationship to Parkinson's disease (PD), and rapid progression have sparked interest in drug development. One significant obstacle to therapeutics is disease heterogeneity. Here, we share our process of developing a clinical trial-ready cohort of MSA patients (69 patients in 2 years) within an outpatient clinical setting, and recruiting 20 of these patients into a longitudinal "n-of-few" clinical trial paradigm. First, we deeply phenotype our patients with clinical scales (UMSARS, BARS, MoCA, NMSS, and UPSIT) and tests designed to establish early differential diagnosis (including volumetric MRI, FDG-PET, MIBG scan, polysomnography, genetic testing, autonomic function tests, skin biopsy) or disease activity (PBR06-TSPO). Second, we longitudinally collect biospecimens (blood, CSF, stool) and clinical, biometric, and imaging data to generate antecedent disease-progression scores. Third, in our Mass General Brigham SCiN study (stem cells in neurodegeneration), we generate induced pluripotent stem cell (iPSC) models from our patients, matched to biospecimens, including postmortem brain. We present 38 iPSC lines derived from MSA patients and relevant disease controls (spinocerebellar ataxia and PD, including alpha-synuclein triplication cases), 22 matched to whole-genome sequenced postmortem brain. iPSC models may facilitate matching patients to appropriate therapies, particularly in heterogeneous diseases for which patient-specific biology may elude animal models. We anticipate that deeply phenotyped and genotyped patient cohorts matched to cellular models will increase the likelihood of success in clinical trials for MSA.

A New Framework of Designing Iterative Techniques for Image Deblurring.

In this work we present a framework of designing iterative techniques for image deblurring in inverse problem. The new framework is based on two observations about existing methods. We used Landweber method as the basis to develop and present the new framework but note that the framework is applicable to other iterative techniques. First, we observed that the iterative steps of Landweber method consist of a constant term, which is a low-pass filtered version of the already blurry observation. We proposed a modification to use the observed image directly. Second, we observed that Landweber method uses an estimate of the true image as the starting point. This estimate, however, does not get updated over iterations. We proposed a modification that updates this estimate as the iterative process progresses. We integrated the two modifications into one framework of iteratively deblurring images. Finally, we tested the new method and compared its performance with several existing techniques, including Landweber method, Van Cittert method, GMRES (generalized minimal residual method), and LSQR (least square), to demonstrate its superior performance in image deblurring.

Deep-Learning Detection of Cancer Metastases to the Brain on MRI.

BACKGROUND: Approximately one-fourth of all cancer metastases are found in the brain. MRI is the primary technique for detection of brain metastasis, planning of radiotherapy, and the monitoring of treatment response. Progress in tumor treatment now requires detection of new or growing metastases at the small subcentimeter size, when these therapies are most effective. PURPOSE: To develop a deep-learning-based approach for finding brain metastasis on MRI. STUDY TYPE: Retrospective. SEQUENCE: Axial postcontrast 3D T1 -weighted imaging. FIELD STRENGTH: 1.5T and 3T. POPULATION: A total of 361 scans of 121 patients were used to train and test the Faster region-based convolutional neural network (Faster R-CNN): 1565 lesions in 270 scans of 73 patients for training; 488 lesions in 91 scans of 48 patients for testing. From the 48 outputs of Faster R-CNN, 212 lesions in 46 scans of 18 patients were used for training the RUSBoost algorithm (MatLab) and 276 lesions in 45 scans of 30 patients for testing. ASSESSMENT: Two radiologists diagnosed and supervised annotation of metastases on brain MRI as ground truth. This data were used to produce a 2-step pipeline consisting of a Faster R-CNN for detecting abnormal hyperintensity that may represent brain metastasis and a RUSBoost classifier to reduce the number of false-positive foci detected. STATISTICAL TESTS: The performance of the algorithm was evaluated by using sensitivity, false-positive rate, and receiver's operating characteristic (ROC) curves. The detection performance was assessed both per-metastases and per-slice. RESULTS: Testing on held-out brain MRI data demonstrated 96% sensitivity and 20 false-positive metastases per scan. The results showed an 87.1% sensitivity and 0.24 false-positive metastases per slice. The area under the ROC curve was 0.79. CONCLUSION: Our results showed that deep-learning-based computer-aided detection (CAD) had the potential of detecting brain metastases with high sensitivity and reasonable specificity. LEVEL OF EVIDENCE: 3 TECHNICAL EFFICACY STAGE: 2 J. Magn. Reson. Imaging 2020;52:1227-1236.

A New Design in Iterative Image Deblurring for Improved Robustness and Performance.

In many applications, image deblurring is a pre-requisite to improve the sharpness of an image before it can be further processed. Iterative methods are widely used for deblurring images but care must be taken to ensure that the iterative process is robust, meaning that the process does not diverge and reaches the solution reasonably fast, two goals that sometimes compete against each other. In practice, it remains challenging to choose parameters for the iterative process to be robust. We propose a new approach consisting of relaxed initialization and pixel-wise updates of the step size for iterative methods to achieve robustness. The first novel design of the approach is to modify the initialization of existing iterative methods to stop a noise term from being propagated throughout the iterative process. The second novel design is the introduction of a vectorized step size that is adaptively determined through the iteration to achieve higher stability and accuracy in the whole iterative process. The vectorized step size aims to update each pixel of an image individually, instead of updating all the pixels by the same factor. In this work, we implemented the above designs based on the Landweber method to test and demonstrate the new approach. Test results showed that the new approach can deblur images from noisy observations and achieve a low mean squared error with a more robust performance.

Normalization of ADC does not improve correlation with overall survival in patients with high-grade glioma (HGG).

Mixed reports leave uncertainty about whether normalization of apparent diffusion coefficient (ADC) to a within-subject white matter reference is necessary for assessment of tumor cellularity. We tested whether normalization improves the previously reported correlation of resection margin ADC with 15-month overall survival (OS) in HGG patients. Spin-echo echo-planar DWI was retrieved from 3 T MRI acquired between maximal resection and radiation in 37 adults with new-onset HGG (25 glioblastoma; 12 anaplastic astrocytoma). ADC maps were produced with the FSL DTIFIT tool (Oxford Centre for Functional MRI). 3 neuroradiologists manually selected regions of interest (ROI) in normal appearing white matter (NAWM) and in non-enhancing tumor (NT) < 2 cm from the margin of residual enhancing tumor or resection cavity. Normalized ADC (nADC) was computed as the ratio of absolute NT ADC to NAWM ADC. Reproducibility of nADC and absolute ADC among the readers' ROI was assessed using intra-class correlation coefficient (ICC) and within-subject coefficient of variation (wCV). Correlations of ADC and nADC with OS were compared using receiver operating characteristics (ROC) analysis. A p value 0.05 was considered statistically significant. Both mean ADC and nADC differed significantly between patients subgrouped by 15-month OS (p = 0.0014 and 0.0073 respectively). wCV and ICC among the readers were similar for absolute and normalized ADC. In ROC analysis of correlation with OS, nADC did not perform significantly better than absolute ADC. Normalization does not significantly improve the correlation of absolute ADC with OS in HGG, suggesting that normalization is not necessary for clinical or research ADC analysis in HGG patients.

Neuroimaging in Pregnant Women.

Choosing the most appropriate diagnostic neuroimaging study for a pregnant woman involves assessing the pretest likelihood of serious treatable neurologic disease, the diagnostic utility of various available computed tomography (CT) and magnetic resonance (MR) modalities, and the risks of each. Of these three elements-pretest differential diagnosis, utility of MRI and CT, and risks of MR and CT-the risk component is perhaps the least well understood by most physicians. We provide a basic review of the intrinsic risks of MRI and CT, particularly radiation biology and radiation safety, as well as the risks pertaining to the use of contrast agents, to reduce provider confusion and anxiety and improve quality, safety, and efficiency of neuroimaging diagnosis in pregnant patients. We believe that a better understanding of the associated very low risks with mother and fetus will reassure the reader that CT remains the most appropriate tool for initial rapid diagnosis of acute neurological conditions in pregnancy and that in urgent situations CT should not be withheld or delayed due to exaggerated concern about radiation. Noncontrast MRI, while not without risk, is generally considered safe in pregnancy, as no evidence of fetal adverse effects has been demonstrated to date. Iodinated CT contrast agents are likely safer than gadolinium-based MRI contrast agents because of gadolinium accumulation in the amniotic fluid and fetal tissue, although no harmful effects of tissue gadolinium accumulation are known. In most but not all pregnant patients presenting with a new or worsening neurological abnormality, the risks intrinsic to the disease will outweigh the risks of imaging. In an individual patient, the pretest probability of serious treatable disease and acuity of presentation will usually suggest an optimal imaging strategy and choice of test. This optimal strategy will also depend on the immediate availability and level of sophistication of the scanners, software, technologists, and radiologists. As such, the standard of care for imaging in pregnancy requires direct consultation between the referring clinician and radiologist to determine the most appropriate strategy and brief documentation of the resulting consensus risk-benefit assessment.

Advanced MRI assessment to predict benefit of anti-programmed cell death 1 protein immunotherapy response in patients with recurrent glioblastoma.

INTRODUCTION: We describe the imaging findings encountered in GBM patients receiving immune checkpoint blockade and assess the potential of quantitative MRI biomarkers to differentiate patients who derive therapeutic benefit from those who do not. METHODS: A retrospective analysis was performed on longitudinal MRIs obtained on recurrent GBM patients enrolled on clinical trials. Among 10 patients with analyzable data, bidirectional diameters were measured on contrast enhanced T1 (pGd-T1WI) and volumes of interest (VOI) representing measurable abnormality suggestive of tumor were selected on pGdT1WI (pGdT1 VOI), FLAIR-T2WI (FLAIR VOI), and ADC maps. Intermediate ADC (IADC) VOI represented voxels within the FLAIR VOI having ADC in the range of highly cellular tumor (0.7-1.1 × 10-3 mm2/s) (IADC VOI). Therapeutic benefit was determined by tissue pathology and survival on trial. IADC VOI, pGdT1 VOI, FLAIR VOI, and RANO assessment results were correlated with patient benefit. RESULTS: Five patients were deemed to have received therapeutic benefit and the other five patients did not. The average time on trial for the benefit group was 194 days, as compared to 81 days for the no benefit group. IADC VOI correlated well with the presence or absence of clinical benefit in 10 patients. Furthermore, pGd VOI, FLAIR VOI, and RANO assessment correlated less well with response. CONCLUSION: MRI reveals an initial increase in volumes of abnormal tissue with contrast enhancement, edema, and intermediate ADC suggesting hypercellularity within the first 0-6 months of immunotherapy. Subsequent stabilization and improvement in IADC VOI appear to better predict ultimate therapeutic benefit from these agents than conventional imaging.

Spin-echo echo-planar perfusion prior to chemoradiation is a strong independent predictor of progression-free and overall survival in newly diagnosed glioblastoma.

Spin-echo echo planar (EP) perfusion weighted imaging (SE-PWI) has been demonstrated to be more selective than gradient-echo EP PWI for blood volume in microvessels the size of glioma neocapillaries, but it has not been comprehensively studied in human clinical use. We assessed whether SE-PWI before and after initiating chemoradiation can stratify patients with respect to progression free survival (PFS) and overall survival (OS). Sixty-eight patients with newly diagnosed glioblastoma (mean age 58.3, 36 males) were included in analysis. SE EP cerebral blood volumes (SE-CBVs) in enhancing and nonenhancing tumor, normalized to contralateral normal appearing white matter (SE-nCBV), were assessed at baseline and after initial chemoradiation. SE-nCBV parameters predictive of PFS and OS were identified in univariate and multivariate Cox proportional hazards models. Multivariate analysis demonstrated that baseline tumor mean SE-nCBV was predictive of PFS (p = 0.038) and OS (p = 0.004). Within the patient sample, baseline tumor mean SE-nCBV <2.0 predicted longer patient PFS (median 47.0 weeks, p < 0.001) and OS (median 98.6 weeks, p = 0.003) compared with baseline mean SE-nCBV >2.0 (median PFS 25.3, median OS 56.0 weeks). Exploratory multi-group stratification demonstrated that very high (>4.0) tumor SE-nCBV was associated with worse patient OS than intermediate high (>2.0, <4.0) SE-nCBV (p = 0.025). Baseline mean SE-nCBV can stratify patients for PFS and OS prior to initiation of chemoradiation, which may help select patients who require closer surveillance. Our exploratory analysis indicates a magnitude-dependent relationship between baseline SE-nCBV and OS.

Auto-segmentation of Adult-Type Diffuse Gliomas: Comparison of Transfer Learning-Based Convolutional Neural Network Model vs. Radiologists.

Segmentation of glioma is crucial for quantitative brain tumor assessment, to guide therapeutic research and clinical management, but very time-consuming. Fully automated tools for the segmentation of multi-sequence MRI are needed. We developed and pretrained a deep learning (DL) model using publicly available datasets A (n = 210) and B (n = 369) containing FLAIR, T2WI, and contrast-enhanced (CE)-T1WI. This was then fine-tuned with our institutional dataset (n = 197) containing ADC, T2WI, and CE-T1WI, manually annotated by radiologists, and split into training (n = 100) and testing (n = 97) sets. The Dice similarity coefficient (DSC) was used to compare model outputs and manual labels. A third independent radiologist assessed segmentation quality on a semi-quantitative 5-scale score. Differences in DSC between new and recurrent gliomas, and between uni or multifocal gliomas were analyzed using the Mann-Whitney test. Semi-quantitative analyses were compared using the chi-square test. We found that there was good agreement between segmentations from the fine-tuned DL model and ground truth manual segmentations (median DSC: 0.729, std-dev: 0.134). DSC was higher for newly diagnosed (0.807) than recurrent (0.698) (p < 0.001), and higher for unifocal (0.747) than multi-focal (0.613) cases (p = 0.001). Semi-quantitative scores of DL and manual segmentation were not significantly different (mean: 3.567 vs. 3.639; 93.8% vs. 97.9% scoring ≥ 3, p = 0.107). In conclusion, the proposed transfer learning DL performed similarly to human radiologists in glioma segmentation on both structural and ADC sequences. Further improvement in segmenting challenging postoperative and multifocal glioma cases is needed.

Multicolored stain-free histopathology with coherent Raman imaging.

Conventional histopathology with hematoxylin & eosin (H&E) has been the gold standard for histopathological diagnosis of a wide range of diseases. However, it is not performed in vivo and requires thin tissue sections obtained after tissue biopsy, which carries risk, particularly in the central nervous system. Here we describe the development of an alternative, multicolored way to visualize tissue in real-time through the use of coherent Raman imaging (CRI), without the use of dyes. CRI relies on intrinsic chemical contrast based on vibrational properties of molecules and intrinsic optical sectioning by nonlinear excitation. We demonstrate that multicolor images originating from CH(2) and CH(3) vibrations of lipids and protein, as well as two-photon absorption of hemoglobin, can be obtained with subcellular resolution from fresh tissue. These stain-free histopathological images show resolutions similar to those obtained by conventional techniques, but do not require tissue fixation, sectioning or staining of the tissue analyzed.

Apparent diffusion coefficient histogram analysis stratifies progression-free and overall survival in patients with recurrent GBM treated with bevacizumab: a multi-center study.

We have tested the predictive value of apparent diffusion coefficient (ADC) histogram analysis in stratifying progression-free survival (PFS) and overall survival (OS) in bevacizumab-treated patients with recurrent glioblastoma multiforme (GBM) from the multi-center BRAIN study. Available MRI's from patients enrolled in the BRAIN study (n = 97) were examined by generating ADC histograms from areas of enhancing tumor on T1 weighted post-contrast images fitted to a two normal distribution mixture curve. ADC classifiers including the mean ADC from the lower curve (ADC-L) and the mean lower curve proportion (LCP) were tested for their ability to stratify PFS and OS by using Cox proportional hazard ratios and the Kaplan-Meier method with log-rank test. Mean ADC-L was 1,209 × 10(-6)mm(2)/s ± 224 (SD), and mean LCP was 0.71 ± 0.23 (SD). Low ADC-L was associated with worse outcome. The hazard ratios for 6-month PFS, overall PFS, and OS in patients with less versus greater than mean ADC-L were 3.1 (95 % confidence interval: 1.6, 6.1; P = 0.001), 2.3 (95 % CI: 1.3, 4.0; P = 0.002), and 2.4 (95 % CI: 1.4, 4.2; P = 0.002), respectively. In patients with ADC-L <1,209 and LCP >0.71 versus ADC-L >1,209 and LCP <0.71, there was a 2.28-fold reduction in the median time to progression, and a 1.42-fold decrease in the median OS. The predictive value of ADC histogram analysis, in which low ADC-L was associated with poor outcome, was confirmed in bevacizumab-treated patients with recurrent GBM in a post hoc analysis from the multi-center (BRAIN) study.

Cerebral perfusion imaging.

Recent advances are allowing computed tomography (CT) and magnetic resonance imaging (MRI) to add diagnostic information derived from microscopic-scale brain structure, pathology, and physiology to the gross pathologic information that has been the core of brain imaging diagnosis since the 1980s. Physiologic imaging with MR perfusion weighted imaging has joined MR diffusion imaging as an essential components of stroke and brain tumor MRI. At the same time, the volume scanning revolution in CT technology has dramatically decreased the radiation doses required for CT perfusion imaging by allowing routine simultaneous CT perfusion and noninvasive dynamic bone subtracted CT angiography (CTA) without a significant increase in radiation dose over conventional head CT-CTA alone. Although ongoing research and clinical trials is needed to define more precisely how these techniques can best be exploited to improve clinical care and patient outcomes, in the acute stroke and subarachnoid hemorrhage populations the radiation risk associated with CT perfusion imaging is negligible and the physiologic information promises significant patient safety benefits.

Clinical 7T MRI for epilepsy care: Value, patient selection, technical issues, and outlook.

Ultra-high-field 7.0 Tesla (T) MRI offers substantial gains in signal-to-noise ratio (SNR) over 3T and 1.5T, but for over two decades has remained a research tool, while 3T scanners have achieved widespread clinical use. This much slower translation of 7T relates to daunting technical challenges encountered in ultra-high-field human MR imaging. The recent introduction of United States Food and Drug Administration (FDA)-approved clinical 7T scanners promises to be a watershed for many 7T neuroimaging applications, including epilepsy imaging. The high SNR of 7T allows clinical imaging of fine neuroanatomic detail at unprecedented spatial resolution, helping with detection and differentiation of subtle, potentially treatable lesions undetectable or suboptimally assessed at 3T. The accompanying research paper reports our group's analysis of 7T MRI efficacy in epilepsy treatment planning. Here, we introduce the technical background and clinical approach we currently use, in order to assist clinical epileptologists and neuroimagers contemplating, creating, or referring patients to a clinical 7T epilepsy imaging service. We describe a tiered epilepsy imaging strategy and protocols designed to optimize 7T value and work around signal intensity variation and signal loss artifacts, which remain significant challenges to full exploitation of 7T clinical value. We describe FDA-approved techniques for mitigating these artifacts and briefly outline techniques currently under development, but not yet FDA approved. Finally, we discuss the major issues in 7T patient safety and toleration, outlining their physical causes and effects on workflow, and provide references to more comprehensive technical reviews for readers seeking greater technical detail.

7T versus 3T MRI in the presurgical evaluation of patients with drug-resistant epilepsy.

BACKGROUND AND PURPOSE: MRI has a crucial role in presurgical evaluation of drug-resistant focal epilepsy patients. Whether and how much 7T MRI further improves presurgical diagnosis compared to standard of care 3T MRI remains to be established. We investigate the added value 7T MRI offers in surgical candidates with remaining clinical uncertainty after 3T MRI. METHODS: 7T brain MRI was obtained on sequential patients with drug-resistant focal epilepsy undergoing presurgical evaluation at a comprehensive epilepsy center, including patients with and without suspected lesions on standard 3T MRI. Clinical information and 3T images informed the interpretation of 7T images. Detection of a new lesion on 7T or better characterization of a suspected lesion was considered to add value to the presurgical workup. RESULTS: Interpretable 7T MRI was acquired in 19 patients. 7T MRI identified a lesion relevant to the seizures in three of eight patients (38%) without a lesion on 3T MRI; no lesion in 7/11 patients (64%) with at least one suspected lesion on 3T MRI, contributing to the final classification of all seven as nonlesional; and confirmed and better characterized the lesion suspected at 3T MR in the remaining 4/11 patients. CONCLUSIONS: 7T MRI detected new lesions in over a third of 3T MRI nonlesional patients, confirmed and better characterized a 3T suspected lesion in one third of patients, and helped exclude a 3T suspected lesion in the remainder. Our initial experience suggests that 7T MRI adds value to surgical planning by improving detection and characterization of suspected brain lesions in drug-resistant focal epilepsy patients.

Support vector machine multiparametric MRI identification of pseudoprogression from tumor recurrence in patients with resected glioblastoma.

PURPOSE: To automatically differentiate radiation necrosis from recurrent tumor at high spatial resolution using multiparametric MRI features. MATERIALS AND METHODS: MRI data retrieved from 31 patients (15 recurrent tumor and 16 radiation necrosis) who underwent chemoradiation therapy after surgical resection included post-gadolinium T1, T2, fluid-attenuated inversion recovery, proton density, apparent diffusion coefficient (ADC), and perfusion-weighted imaging (PWI) -derived relative cerebral blood volume (rCBV), relative cerebral blood flow (rCBF), and mean transit time maps. After alignment to post contrast T1WI, an eight-dimensional feature vector was constructed. An one-class-support vector machine classifier was trained using a radiation necrosis training set. Classifier parameters were optimized based on the area under receiver operating characteristic (ROC) curve. The classifier was then tested on the full dataset. RESULTS: The sensitivity and specificity of optimized classifier for pseudoprogression was 89.91% and 93.72%, respectively. The area under ROC curve was 0.9439. The distribution of voxels classified as radiation necrosis was supported by the clinical interpretation of follow-up scans for both nonprogressing and progressing test cases. The ADC map derived from diffusion-weighted imaging and rCBV, rCBF derived from PWI were found to make a greater contribution to the discrimination than the conventional images. CONCLUSION: Machine learning using multiparametric MRI features may be a promising approach to identify the distribution of radiation necrosis tissue in resected glioblastoma multiforme patients undergoing chemoradiation.

Characterization of a human tumorsphere glioma orthotopic model using magnetic resonance imaging.

Magnetic resonance imaging (MRI) is the imaging modality of choice by which to monitor patient gliomas and treatment effects, and has been applied to murine models of glioma. However, a major obstacle to the development of effective glioma therapeutics has been that widely used animal models of glioma have not accurately recapitulated the morphological heterogeneity and invasive nature of this very lethal human cancer. This deficiency is being alleviated somewhat as more representative models are being developed, but there is still a clear need for relevant yet practical models that are well-characterized in terms of their MRI features. Hence we sought to chronicle the MRI profile of a recently developed, comparatively straightforward human tumor stem cell (hTSC) derived glioma model in mice using conventional MRI methods. This model reproduces the salient features of gliomas in humans, including florid neoangiogenesis and aggressive invasion of normal brain. Accordingly, the variable, invasive morphology of hTSC gliomas visualized on MRI duplicated that seen in patients, and it differed considerably from the widely used U87 glioma model that does not invade normal brain. After several weeks of tumor growth the hTSC model exhibited an MRI contrast enhancing phenotype having variable intensity and an irregular shape, which mimicked the heterogeneous appearance observed with human glioma patients. The MRI findings reported here support the use of the hTSC glioma xenograft model combined with MRI, as a test platform for assessing candidate therapeutics for glioma, and for developing novel MR methods.

Longitudinal MRI evidence for decreased survival among periventricular glioblastoma.

While the prognosis of patients with glioblastoma (GBM) remains poor despite recent therapeutic advances, variable survival times suggest wide variation in tumor biology and an opportunity for stratified intervention. We used volumetric analysis and morphometrics to measure the spatial relationship between subventricular zone (SVZ) proximity and survival in a cohort of 39 newly diagnosed GBM patients. We collected T2-weighted and gadolinium-enhanced T1-weighted magnetic resonance images (MRI) at pre-operative, post-operative, pre-radiation therapy, and post-radiation therapy time points, measured tumor volumes and distances to the SVZ, and collected clinical data. Univariate and multivariate Cox regression showed that tumors involving the SVZ and tumor growth rate during radiation therapy were independent predictors of shorter progression-free and overall survival. These results suggest that GBMs in close proximity to the ependymal surface of the ventricles convey a worse prognosis-an observation that may be useful for stratifying treatment.