Many segmentation and annotation tasks can be automated by machine learning approaches, that are trained on manually labeled examples. We propose methods to reduce the required number of labeled examples by exploiting labels in other domains and by selecting the most relevant images for labeling.
Active & Transfer Learning
Acquiring annotated training data is a major bottleneck in supervised learning, especially in the biomedical domain because it has to be done on 3D data by experts whose time is precious and often after each new image acquisition.
Active and Transfer Learning techniques mitigate this problem and reduce the annotation cost either by allowing only the most informative examples to be annotated or by leveraging annotations from previous image acquistions.