CSBDeep

CSBDeep is a Python toolbox for researchers who want to restore fluorescence microscopy images using content-aware restoration (CARE). It covers training restoration models and applying them to images, with deep learning built on Keras and TensorFlow.

The package includes several CARE model types. Alongside the standard restoration model, it provides UpsamplingCARE for upsampling, IsotropicCARE for isotropic restoration, and ProjectionCARE for projection tasks. These give microscopy researchers model choices for distinct image restoration problems within the same package.

Its scope also includes preparing training data. The toolbox supports anisotropic distortions, data augmentation and other image transforms, while image preparation includes normalization and resizing. Researchers can use these capabilities as part of a workflow that covers data preparation, model training and model application.

CSBDeep is open source under the BSD-3-Clause license. It includes Jupyter notebook examples that demonstrate how to use the package, and the documentation describes a GPU-enabled installation option for Linux.