Noise2Void
Noise2Void is a deep-learning method that can be used to denoise microscopy images. No specific training datasets are required, only your noisy images. One noisy image is sufficient to train a network.
This page contains information to help you train Noise2Void networks in Google Colab using your own images.
Noise2Void was described by Krull et al. Learning Denoising from Single Noisy Images
Noise2Void original code and documentation are freely available in GitHub.
Please also cite this original paper when training Noise2Void with our notebooks.
To train a Noise2Void network, all you need are your noisy images. One noisy image is even sufficient to train a network.
The dataset provided as an example with our notebooks was generated by plating U-251 glioma cells expressing endogenously tagged paxillin-GFP on fibronectin-coated polyacrylamide gels (stiffness 9.6 Kpa) (Stubb et al, 2020). Cells were then recorded live using a spinning disk confocal microscope equipped with a long working distance 63x (NA 1.15 water, LD C-Apochromat) objective (Zeiss). The spinning disk confocal microscope used was a Marianas spinning disk imaging system with a Yokogawa CSU-W1 scanning unit on an inverted Zeiss Axio Observer Z1 microscope controlled by SlideBook 6 (Intelligent Imaging Innovations, Inc.). Images were acquired using a Photometrics Evolve, a back-illuminated EMCCD camera (512 x 512 pixels).
To train Noise2Void in Google Colab:
| Network | Link to example training and test dataset | Direct link to notebook in Colab |
|---|---|---|
| Noise2Void (2D) | here |  |
| Noise2Void (3D) | here | |
or:
-
Download our streamlined ZeroCostDL4Mic notebooks
-
Open Google Colab
-
Once the notebook is open, follow the instructions.