Learning with minimal effort: leveraging in silico labeling for cell and nucleus segmentation
This addresses the tedious and expensive annotation process for biomedical imaging, though it is incremental as it builds on existing ISL methods.
The paper tackles the problem of reducing manual annotation effort for cell and nucleus segmentation by proposing In Silico Labeling (ISL) as a pretraining scheme, showing it dramatically reduces the number of required annotations.
Deep learning provides us with powerful methods to perform nucleus or cell segmentation with unprecedented quality. However, these methods usually require large training sets of manually annotated images, which are tedious and expensive to generate. In this paper we propose to use In Silico Labeling (ISL) as a pretraining scheme for segmentation tasks. The strategy is to acquire label-free microscopy images (such as bright-field or phase contrast) along fluorescently labeled images (such as DAPI or CellMask). We then train a model to predict the fluorescently labeled images from the label-free microscopy images. By comparing segmentation performance across several training set sizes, we show that such a scheme can dramatically reduce the number of required annotations.