Faculty Bibliography

PURPOSE/OBJECTIVE:Recent advances in sodium brain MRI have allowed for increased signal-to-noise ratio, faster imaging, and the ability of differentiating intracellular from extracellular sodium concentration, opening a new window of opportunity for clinical application. In gliomas, there are significant alterations in sodium metabolism, including increase in the total sodium concentration and extracellular volume fraction. The purpose of this study is to assess the feasibility of using sodium MRI quantitative measurements to evaluate gliomas. METHODS:), apparent intracellular sodium concentration (aISC), and apparent total sodium concentration (aTSC). Measurements were made within the contralateral normal-appearing putamen, contralateral normal-appearing white matter (NAWM), and solid tumor regions (area of T2-FLAIR abnormality, excluding highly likely areas of edema, cysts, or necrosis). Paired samples t test were performed comparing NAWM and putamen and between NAWM and solid tumor. RESULTS:(p = 0.19). CONCLUSION/CONCLUSIONS:Quantitative sodium measurements can be done in glioma patients and also has provided further evidence that total sodium and extracellular volume fraction are increased in gliomas.

BACKGROUND AND PURPOSE/OBJECTIVE:The World Health Organization has recently placed new emphasis on the integration of genetic information for gliomas. While tissue sampling remains the criterion standard, noninvasive imaging techniques may provide complimentary insight into clinically relevant genetic mutations. Our aim was to train a convolutional neural network to independently predict underlying molecular genetic mutation status in gliomas with high accuracy and identify the most predictive imaging features for each mutation. MATERIALS AND METHODS/METHODS:) promotor methylation status. Principal component analysis of the final convolutional neural network layer was used to extract the key imaging features critical for successful classification. RESULTS:promotor methylation status, 83%. Each genetic category was also associated with distinctive imaging features such as definition of tumor margins, T1 and FLAIR suppression, extent of edema, extent of necrosis, and textural features. CONCLUSIONS:Our results indicate that for The Cancer Imaging Archives dataset, machine-learning approaches allow classification of individual genetic mutations of both low- and high-grade gliomas. We show that relevant MR imaging features acquired from an added dimensionality-reduction technique demonstrate that neural networks are capable of learning key imaging components without prior feature selection or human-directed training.

Physiological and pathological processes that increase or decrease the central nervous system's need for nutrients and oxygen via changes in local blood supply act primarily at the level of the neurovascular unit (NVU). The NVU consists of endothelial cells, associated blood-brain barrier tight junctions, basal lamina, pericytes, and parenchymal cells, including astrocytes, neurons, and interneurons. Knowledge of the NVU is essential for interpretation of central nervous system physiology and pathology as revealed by conventional and advanced imaging techniques. This article reviews current strategies for interrogating the NVU, focusing on vascular permeability, blood volume, and functional imaging, as assessed by ferumoxytol an iron oxide nanoparticle.

Purpose: Prior studies have shown correlation between relative cerebral blood volume (rCBV) and patient survival and tumor genomics. The purpose of this study was to determine whether rCBV values correlate with isocitrate dehydrogenase (IDH) mutation status, genome-wide CNV (copy number variation) and patient overall survival in diffuse gliomas. Materials & Methods: 107 treatment naive gliomas (62 patients from TCGA/TCIA dataset and 45 patients from our institute) (44 glioblastoma and 63 lower grade gliomas) with DSC T2* perfusion data were included. IDH mutation and survival data were assayed by the TCGA, and pre-surgical imaging collected by The Cancer Imaging Archive. CNVabundance plots obtained with Illumina 850k EPIC DNA methylation arrays were reviewed in 19 patients. The association of rCBV with tumor genomics, CNV and overall survival were analyzed. Results: IDH-wildtype gliomas (44.8%) demonstrated higher rCBV values (rCBV = 6.87 +/- 3.09) than IDH-mutated gliomas (55.2%, rCBV =2.21 +/- 1.71 for 1p/19q codeleted gliomas and 2.09 +/- 2.00 for non-codeleted gliomas, ANOVA, p<0.0001). rCBV is a significant predictor of overall survival (HR 1.23, p<0.0001). Gliomas with rCBV < 3.80 showed better survival (n = 54, median survival time unobserved) than gliomas with rCBV > 3.8 (n = 53, median 18 months; log-rank p<0.0001). IDHwt gliomas with high rCBV had the worst survival (10.6% surviving at 3 years, 95% CI (4%, 30%)). CNV-S IDHmut 1p19q noncodeleted gliomas demonstrated significantly lower mean rCBV (1.4 +/- 0.4) than CNV-U gliomas (4.0 +/- 1.1, p = 0.009). Conclusion: IDHwt gliomas show higher rCBV than IDHmut gliomas irrespective of the glioma grade. Higher rCBV measurements are associated with poorer survival in the entire cohort and also within IDHmut and IDHwt gliomas. IDHmut 1p19q noncodeleted gliomas with higher CNV abundance (CNV-U) also show higher CBV when compared with those with lesser degree of CNVabundance (CNV-S)

BACKGROUND: Resting state fMRI (rsfMRI) is task independent mapping of individual brain function. Here we demonstrated the clinical feasibility of rsfMRI in patients with brain tumor. MATERIALS AND METHODS: Resting-state fMRI data from a total of 39 cases (32 primary brain tumor patients [8 tumors in sensorimotor cortices] and 7 normal healthy control cases) were analyzed based on independent component analysis (ICA). For each patient, ipsilesional(IL) and contralesional(CL) regions of interest (ROI) from supra-threshold Z-score voxels (Z-score > 2.3) on sensorimotor network (SMN) were used to evaluate BOLD signal changes on Z score maps. Asymmetry score were also calculated based on numbers of supra-thershold voxels between IL and CL in tumor patients and between I and C ROIs in control cases. Statistically significant differences between the each subgroup were tested using a two-tailed t-test. RESULTS: In patients with tumor involving SMN, overall decreased BOLD signal by supra-threshold voxel count and mean Z-score was noted in both IL and CL ROIs (mean +/- standard deviation, voxel number: IL: 2376 +/- 1366, CL: 2906 +/- 1210, mean Z-score: IL: 3.7 +/- 0.3, CL: 3.65 +/- 0.3) compared to brain tumor located outside SMN (voxel number: IL: 3258 +/- 1370, CL: 3413 +/- 1467, mean Zscore: IL: 4.0 +/- 0.4, CL: 3.9 +/- 0.4). The asymmetry score was also higher in tumor involving SMN group than those not (0.1 vs 0.02). Overall increased amplitude of BOLD signal and decreased spatial extent of network of whole SMN were depicted in patients with brain tumor as compared with normal subjects. However, these differences were not statistically significant. CONCLUSION: We have demonstrated that brain tumor might cause changes of the spatial extent and amplitude of SMN on rsfRMI without significant differences observed. Resting fMRI and connectivity analysis are task-independent, and have potential clinical utility in the presurgical evaluation of patients with brain tumors, and may help in uncooperative or pediatric patients