Abstract #2779

Automatic Segmentation and Classification of Glioblastoma using DCE-MRI

Moran Artzi¹, Gilad Liberman^2,3, Deborah Blumenthal⁴, Felix Bokstein⁴, Orna Aizenstein¹, and Dafna Ben Bashat^1,5

¹Functional Brain Center, Tel Aviv Sourasky Medical Center, Tel-Aviv, Israel, ²Functional Brain Centerasky Medical Cente, Tel Aviv Sourasky Medical Center, Tel-Aviv, Israel, ³Department of Chemical Physics, Weizmann Institute, Rehovot, Israel, ⁴Neuro-Oncology Service, Tel Aviv Sourasky Medical Center, Tel-Aviv, Israel, ⁵Sackler Faculty of Medicine and Sagol School of Neuroscience, Tel Aviv University, Tel-Aviv, Israel

Segmentation of lesion area in patients with glioblastoma (GB) into active tumor, tissue necrosis, vasogenic edema and infiltrative disease, is highly important for patient monitoring, yet is challenging using standard radiological assessment. The aim of this study was to segment the lesion area into these four tissue types in GB patients. Voxel-wise classification was performed using support-vector-machine based on anatomical and DCE-MRI parameters. Significant differences were detected between the tissue types for FLAIR, v_p, and k^trans. Sensitivity and specificity of the training-set were measured based on 2-fold-cross-validation analysis, showing high sensitivities and specificities of 94-100% for the different tissue types.

This abstract and the presentation materials are available to members only; a login is required.

Join Here