Manual labeling strategy for ground truth estimation in MRI glial tumor segmentation

In this paper we focused our attention on the problem of determining reliable ground truth for validating unsupervised, fully automatic MRI brain tumor segmentation procedures in the clinical context of Glial Tumor treatment. The goal was achieved by proposing an integrated "visual knowledge elicitation strategy" centered on the use of GliMAn(Glial Tumor Manual Annotator), a 3D MRI navigator that allows to view and manually labeling MRI volumes. As seen in our experimental context, the manual labeling process benefits from the insertion of a software tool taylored on the experts visual and usability requirements.