mcmicro module for training and processing large images with Ilastik
Script for preparing ome.tif images to be accessed by Ilastik. Exports hdf5 formats.
Headless Ilastik execution once the classifier is ready
mkdir prob_maps/ilastik
python CommandIlastikPrepOME.py --input exemplar-001.ome.tif --output prob_maps/ilastik/ --num_channels 12
To apply an existing classifier to an hdf5 file created in the previous step:
/path/to/ilastik/run_ilastik.sh --headless --project=classifiers/exemplar_001.ilp prob_maps/ilastik/exemplar-001.hdf5
For training follow these steps
| Parameter | Default | Description |
|---|---|---|
--input |
None |
Path to images (Ex: ./my_image.ome.tif ./my_image2.ome.tif) |
--output |
None |
Path to output directory. Either single directory or number of directories=to number of images (Ex: ./my_outdir) |
--nonzero_fraction <value> |
None |
Indicates fraction of pixels per crop above global threshold to ensure tissue and not only background is selected |
--nuclei_index <index> |
1 |
(1-based) Index of nuclei channel to use for nonzero_fraction argument |
--crop |
Omitted | If specified, crop regions for ilastik training |
--num_channels <value> |
None |
Number of channels to export per image (Ex: 40 corresponds to a 40 channel ome.tif image) |
--channelIDs <indices> |
None |
Integer (1-based) indices specifying which channels to export (Ex: 1 2 4) |
--ring_mask |
Omitted | Specify if you have a ring mask in the same directory to use for reducing size of hdf5 image |
--crop_amount <integer> |
None |
Number of crops you would like to extract |
For example: python CommandIlastikPrepOME.py --input /Users/joshuahess/Desktop/VV_40c.ome.tif /Users/joshuahess/Desktop/VV_40c_test.ome.tif --output /Users/joshuahess/Desktop/TestingIlastik --nonzero_fraction 0.5 --nuclei_index 1 --crop --crop_size 400 400 --num_channels 3 --channelIDs 1 2 3 --ring_mask --crop_amount 2