Skip to main content
Fig. 2 | Cancer Imaging

Fig. 2

From: Radiomics for glioblastoma survival analysis in pre-operative MRI: exploring feature robustness, class boundaries, and machine learning techniques

Fig. 2

Perturbations applied to the single-center data to simulate expected multi-center data quality. Top row: left: Inter-rater simulation by deforming the labels from the automated tumor segmentation; middle: Additive Gaussian noise to match SNR range, measured on a healthy white matter segmentation; right: Adjusting the bin width within the range in the PyRadiomics documentation. Bottom row: K-space subsampling (left original, right subsampled, contrast increased for both for visualization); middle: Voxel size changed isotropically; right: Variations in axial slice spacing

Back to article page

Cancer Imaging

ISSN: 1470-7330

Contact us