Electronic Posters: Neuroimaging - ismrm

Wednesday 13:30-15:30 Computer 74
13:30 4286. Support Vector Machines in DSC-Based Glioma Imaging – Suggestions for Optimal
Characterization
Frank G. Zöllner 1 , Kyrre Eeg Emblem 2 , Lothar R. Schad 1
1 Heidelberg University, Mannheim, Germany; 2 Oslo University Hospital, Oslo, Norway
Dynamic susceptibility contrast magnetic resonance perfusion imaging (DSC-MRI) is a method of choice to characterize gliomas.
Recently, support vector machines (SVM) have been introduced as means to prospectively characterize new patients based on
information from previous patients. Based on features derived from automatically segmented tumor volumes from 101 DSC-MR
examinations, four different SVM models were compared. All SVM models achieved high prediction accuracies (>82%) after
rebalancing the training data sets to equal amounts of samples per class. Best discrimination were obtained using a SVM model with a
radial basis function kernel allowing for a correct prediction of low-grade glioma at 83% and high-grade glioma at 91%.
14:00 4287. Differentiation Between Glioblastomas, Brain Metastases and Primary Cerebral
Lymphomas Using Diffusion and Perfusion Weighted Imaging
Sumei Wang 1 , Sungheon Kim 2 , Sanjeev Chawla 1 , Ronald L. Wolf 1 , David Knipp 1 , Arastoo
Vossough 1 , Donald M. O’Rourke 3 , Kevin D. Judy 3 , Elias R. Melhem 1 , Harish Poptani 1
1 Radiology, University of Pennsylvania, Philadelphia, PA, United States; 2 Radiology, New York University
School of Medicine, New York, NY, United States; 3 Neurosurgery, University of Pennsylvania, Philadelphia,
PA, United States
Twenty-six glioblastomas, 25 brain metastases and 16 cerebral lymphomas underwent DTI and DSC studies. FA, ADC, CL, CP, CS
and rCBV were measured from the enhancing part of the tumor. Elevated FA, CL, CP and decreased CS were observed in
glioblastomas compared with both metastases and lymphomas, whereas ADC and rCBV values from glioblastomas were significantly
higher than lymphomas. FA and ADC was the best predictor for differentiation of glioblastomas from non-glioblastomas, whereas
ADC, CS and rCBV were the best model for distinguishing lymphomas from metastases. Our study indicates that DTI metrics along
with rCBV measurement may be helpful in tumor classification.
14:30 4288. Initial RCBV Predicts Response to Bevacizumab in Patients with High-Grade
Gliomas
Kathleen M. Schmainda 1,2 , Devyani Bedekar 1,2 , Scott D. Rand 1,2 , Jennifer Connelly, 23 ,
Mark Malkin, 23
1 Radiology, Medical College of Wisconsin, Milwaukee, WI, United States; 2 Translational Brain Tumor
Research Program, Medical College of Wisconsin, Milwaukee, WI, United States; 3 Neurology, Medical College
of Wisconsin, Milwaukee, WI, United States
The prognosis for patients diagnosed with brain tumors has been dismal. Now there is hope with improved time to progression and
survival noted for patients treated with anti-angiogenic drugs such as bevacizumab. Yet, many questions remain regarding the
appropriate selection of patients most likely to respond. These questions are important from both a clinical and economic perspective
and therefore speak to the need for ways to efficiently and reliably predict response. Here we demonstrate the potential of DSCmeasures
of rCBV, obtained just prior to treatment with bevacizumab, to predict outcomes in patients with both new and recurrent
GBM.
15:00 4289. Using Cerebrovascular Response to Hyperoxia for Assessing Treatment Resonse in
Glioblastoma
Heisoog Kim 1,2 , Ciprian Catana 1 , Grace Kim 1 , Ovidiu C. Andronesi 1 , Dominique L.
Jennings 1 , Divya S. Bolar 1,3 , Elizabeth R. Gerstner 4 , Tracy T. Batchelor 4 , Rakesh K.
Jain 5 , Alma Gregory Sorensen 1
1 A.A.Martinos center, Massachusetts General Hospotal, Charlestown, MA, United States; 2 NSE/HST,
Massachusetts Institute of Technology, Cambridge, MA, United States; 3 EECS/HST, Massachusetts Institute of
Technology, Cambridge, MA, United States; 4 Neurology, Massachusetts General Hospotal, Boston, MA,
United States; 5 Radiology, Massachusetts General Hospotal, Boston, MA, United States
This study quantitatively investigated BOLD responses to pure oxygen in glioblastoma (GBM) throughout the course of the treatment
with chemoradiation and an anti-angiogenic drug. BOLD signal changes dropped significantly at the beginning of the treatment in
tumor and gradually recovered afterwards. Conversely, in contralateral normal tissue a slight increase was observed at the early time
points. Interestingly, no difference was observed between values in both regions after 35 days. Our preliminary findings suggest that
assessing the oxygenation status before and after treatment might be useful for both prognostic and diagnostic assessment in GBM
patients.