Reducing manual annotation costs for cell segmentation by upgrading low-quality annotations

Deep learning algorithms for image segmentation typically require large data sets with high-quality annotations to be trained with. For many domains, the annotation cost for obtaining such sets may prove to be prohibitively expensive. Our work aims to reduce the time necessary to create high-quality annotated images by using a relatively small well-annotated data set for training a convolutional neural network to upgrade lower-quality annotations, produced at lower annotation costs. We apply our method to the task of cell segmentation and investigate the performance of our solution when upgrading annotation quality for labels affected by three types of annotation errors: omission, inclusion, and bias. We observe that our method is able to upgrade annotations affected by high error levels from 0.3 to 0.9 Dice similarity with the ground-truth annotations. Moreover, we show that a relatively small well-annotated set enlarged with samples with upgraded annotations can be used to train better-performing segmentation networks compared to training only on the well-annotated set.

• View at Publisher