AI framework for perform deconvolution on 3D-volumes:
- create a pre-trained generative adversaraial network (GAN)
- in detail this GAN is an vol2vol (Volume-to-Volume) neuronal network
- train a generator which is able to perform deconvolution directly
- use generator and tailer-made discriminator for an iterative deconvolution model implemented with tensorflow
- Import volumes and specify hyperparameter via the json
- Perform Deconvolution in corresponding jupyter notebook