Home

DeepBugs

Analyzing bacterial bioimages with open-source deep learning approaches

Deep learning is a powerful tool in many areas. In the last years, it also gained large popularity in the field of bioimage analysis. Yet, bacteriologists make only limited use of this technology, although there is great potential.

To leverage the use of DL in microbiology, we created several datasets of bacterial bioimages and tested DL networks for different image analysis tasks.

Our work is based on the ZeroCostDL4Mic platform, as it provides simple access to different DL networks and gives you free GPU computing power! For this it employs the Google Colab platform.

Here's an overview on the tasks that we perform using open-source DL approaches:

Scale is 2 µm

Notebooks used in our work:

All notebooks are implementations provided by the ZeroCostDL4Mic platform. Further documentation can be found on the ZeroCostDL4Mic wiki)

Network	Paper(s)	Task	Link to example training and test dataset	Direct link to notebook in Colab
U-Net (2D)	here and here	Segmentation	ISBI challenge or here
StarDist (2D)	here and here	Nuclei segmentation	here
SplineDist	[here][SplineDist_link]	Instance segmentation	Coming soon!
Noise2Void (2D)	here	Denoising	here
CARE (2D)	here	Denoising	here
Label-free prediction (fnet) 2D	here	Artificial labelling	here
pix2pix	here	Paired Image-to-Image Translation	here
YOLOv2	here	Object detection (bounding boxes)	here

Tools

Network	Paper(s)	Task	Link to example training and test dataset	Direct link to the notebook in Colab
Augmentor	here	Image augmentation	None
Quality Control	Available soon	Error mapping and quality metrics estimation	None

Contributors

• Christoph Spahn
• Romain F. Laine
• Pedro Matos Pereira
• Lucas von Chamier
• Seamus Holden
• Mia Conduit
• Guillaume Jacquemet
• Mike Heilemann
• Ricardo Henriques

You can find our preprint here ## add preprint link here