Fig. 5: Visualization of ViT attentions on the scan.
From: 3D foundation model for generalizable disease detection in head computed tomography
We visualized the focused regions of our model through attention maps, where results show that it can successfully attend to correct disease locations in the majority of cases.
