LMU is using an Airflow server for automating routine tasks related to user management, data archival etc. A side-development is that the server can be used also for setting up automated image analysis workflows.

• Data is uploaded to puhti.csc.fi and analysed.
• Results are downloaded to lmu_active1\airflow.
• Results are kept in lmu_active1\airflow for 2 months.
  • If you find the results useful, please copy them elsewhere.

ImageXpress Micro

The data format of Micro is fixed, which makes it an easy target for automation. To opt-in for the automated analysis, add a file with the name "stardist.cfg" in your folder in L:\lmu_active1\instruments\Micro, for example
L:\lmu_active1\instruments\Micro\hajaalin\stardist.cfg. After that:

• Set the "Storage Location" on MetaXpress as lmu_active1\instruments\Micro\yourusername
• Image your plate as usual.
• Wait for some time (~1-2h).
• Find results in [lmu_active1,lmu_active2]\airflow\yourusername

Nuclei segmentation with StarDist

StarDist is run with default values. Even such it can often produce a segmentation that looks better than what you can quickly get from CellProfiler's IdentifyPrimaryObjects module.

• Results are in lmu_active1\airflow\yourusername\stardist.
• To check if segmentation worked, you can use a modified Napari viewer.
• Here is an example of how you can use the labels in CellProfiler: lmu_active2\image_analysis_examples\cellprofiler
• If you use the labels, remember to cite StarDist. See the bottom of the page.

Image quality measurements with CellProfiler

Here we use CellProfilers measureimagequality module to calculate a set of measurements that in theory could be used to spot images that are out of focus or contain bright, overexposed artifacts. These can cause problems in down-stream analysis, so it would be nice to be able to remove them from the data set. In practice we haven't yet been able to do this reliably without loosing also good images.

Under construction: Illumination correction