Examples

Based on pycromanager

If you are used to Micro-Magellan acquisitions, then the actuator implementation using pycromanagers Acquisition class to start a remote Micro-Magellan acquisition might be interesting for you:

>>> eda_plugin.examples.pyro()

If you have tried the basic example before, please restart Micro-Manager, as Micro-Magellan does not work together well with the PythonEventServer used before.For this implementation no calibration is necessary as described in the basic example. However, all settings have to be done in the Micro-Magellan GUI: Plugins -> Micro-Magellan.

CUDA

If you have CUDA, cuDNN and tensorflow installed you can run the analyser that uses the neural network for image analysis as described in the article.

>>> eda_plugin.examples.keras()

To get an image that the neural network can process, use the ImageInjector plugin installed during the setup process in the on-the-fly processor pipeline in Micro-Manager. (Plugins -> On-the-fly image processing -> ImageInjector) Attention: The ImageInjector has to be deactivated during calibration.

You can download an example dataset by

>>> eda_plugin.download_data("C:/yourpath")

They are tif-stacks of mitochondria and Drp1. For this dataset, Micro-Manager should be set to 2 channels and no Z slices.

As this only uses a different Analyser, it can just as well be used with the pycro-manager based actuator:

>>> eda.examples.pyro_keras()

In this example the actuator is modified as described in Custom Actuators to allow for image injection, as the Micro-Magellan plugins don’t work with the on-the-fly processors. For image injection, the demo camera has to be set to the pixel size of the input tif file.

Tensorflow installation

These guides can help for tensorflow installation:

• Tensorflow install

• Compatibility

• Towardsdatascience blog post

For more advanced things, have a look at Custom Loop Compilation