Abstract #4708

Computer-Aided GBM Tumor ROI Contouring and Mapping Framework for Structural and Physiological MR Images in a Multi-Center Clinical Trial

Jing Huo¹, Matthew S. Brown¹, Jeffry R. Alger^2,3, Whitney B. Pope³, Hyun J. Kim³, Jonathan G. Goldin³

¹Department of Radiological Sciences , UCLA, Los Angeles, CA, USA; ²Department of Neurology, UCLA, Los Angeles, CA, USA; ³Department of Radiological Sciences, UCLA, Los Angeles, CA, USA

The main contribution is to develop a computer-aided 3D multi-modal tool for clinical trials with GBM brain tumor. First we developed a 3D semi-automated segmentation tool to contour the tumor on T1-weighted contrast-enhanced volumetric images and then a 3D tumor ROI mapping tool to map the tumor contours from T1w images onto all the physiological images including dynamic contrast enhance (DCE) images, dynamic susceptibility contrast-enhanced (DSC) and diffusion-weighted (DW) images. We applied the tool to about 650 scans and did the quantitative evaluation with a sub-group of 31 patients. 20 out of 31(64.5%) have over 90% overlap ratio.