Deep feature transfer between localization and segmentation tasks
This approach could alleviate the burden of delineating complex structures in medical images by leveraging easier-to-acquire annotations, though it appears incremental as it builds on existing U-net architectures.
The paper tackles the problem of reducing annotation burden in medical image segmentation by proposing a pre-training scheme where the encoding arm of a U-net is first trained for localization to predict target centers, then transferred to segmentation tasks, applied to optic disc segmentation in fundus photographs.
In this paper, we propose a new pre-training scheme for U-net based image segmentation. We first train the encoding arm as a localization network to predict the center of the target, before extending it into a U-net architecture for segmentation. We apply our proposed method to the problem of segmenting the optic disc from fundus photographs. Our work shows that the features learned by encoding arm can be transferred to the segmentation network to reduce the annotation burden. We propose that an approach could have broad utility for medical image segmentation, and alleviate the burden of delineating complex structures by pre-training on annotations that are much easier to acquire.