This study proposes a radiomics patch-based analysis, based on conventional MRI, for classification of the non-enhancing lesion area into vasogenic edema and infiltrative tumor in patients with high-grade-gliomas. 179 MRI scans obtained from 102 patients were included: 67 patients with high-grade-gliomas and 35 patients with brain-metastases. A total of 225 histogram and gray-level-co-occurrence-matrix based features were extracted from the non-enhancing lesion. Classification was performed using various machine-learning classifiers. The best results were obtained using Linear support-vector-machine, with accuracy=87%, sensitivity=86%, and specificity=89%. Preliminary results in patients treated with bevacizumab demonstrate the clinical potential of this method to improve therapy response assessment.
This abstract and the presentation materials are available to members only; a login is required.