This approach might be preferred if CellProfiler alone is deemed sufficient to achieve a reliable segmentation mask. In this case, the -ilastik_training_ilp parameter doesnt have to be provided, and images that have been pre-processed through ilastik_stack_preprocessing.cppipe will be directly passed to the segmentation.cppipe, therefore bypassing Ilastik. The pipeline has an optional -skip_ilastik parameter which allows the Ilastik pixel classification step to be bypassed in favour of executing a CellProfiler only pipeline. If you would like to follow the naming convention in the schematic above you can name the file ilastik_training_params.ilp.
When exporting, select source as "Probabilities", transpose axis order to "cyx" and export output file as "tiff sequence" to generate the appropriate probability maps. The desired composite RGB images can be provided in tiff format and you can use "membrane", "nuclei" and "background" labels to train your classifier. The input image should be generated by using ilastik_stack_preprocessing.cppipe with CellProfiler.
#CELLPROFILER PIPELINES INSTALL#
The pipeline uses a test-dataset that comes with a pre-trained Ilastik pixel classifier, however if you would like to train a classifier using your own dataset then you will have to download and install Ilastik. If you use a custom module in any of the CellProfiler cppipe files make sure to also add it to the plugins/ directory you will be supplying to the pipeline via the -plugins parameter (see Pipeline execution). If you would like to follow the naming convention in the schematic above you can name the files as either full_stack_preprocessing.cppipe, ilastik_stack_preprocessing.cppipe or segmentation.cppipe. When saving the edited cppipe files, make sure to export the file as a cppipe. Open "CellProfiler > Preferences" and change the "CellProfiler plugins directory" path to where you have stored the custom plugins. You can also create and use you own custom plugins. Custom CellProfiler plugins created by the Bodenmiller group (e.g smoothmultichannel.py and measureobjectintensitymultichannel.py) can be found here. To view, edit and create the cppipe files required by the pipeline you will have to download and install the CellProfiler GUI locally. Other recommended changes to the pipeline are outlined in the Pipeline details section.
See -metadata.csv for the required file format.ĬellProfiler cppipe files with the "NamesAndTypes" module edited to match your antibody panel and desired markers for the identification of cell nuclei and membranes (see Customising inputs).
#CELLPROFILER PIPELINES FULL#
Metadata.csv file containing your antibody panel to identify which tiff files are to be used for the full and Ilastik stacks. Associated antibody panel should contain metal and antibody information in the form of "metal_antibody" e.g. Mcd, tiff or txt data file(s) without any spaces in the file names. However, in order to initially create the custom plugin files required by the pipeline, one needs to install the latest GUI versions of CellProfiler and Ilastik on a local machine (see Customising inputs).
#CELLPROFILER PIPELINES SOFTWARE#
This pipeline is designed to run on most compute infrastructures without the need to pre-install any of the software packages. A more refined and comprehensive pipeline will be uploaded in due course. The project files supplied with the pipeline constitute the minimal requirements to generate a single cell mask. The concept of this step-wise image segmentation by combining Ilastik with CellProfiler was based on the analysis pipeline as described by the Bodenmiller group (Zanotelli & Bodenmiller, Jan 2019). The various stages of this pipeline allow the tiff images to be pre-processed, and segmented using multiple CellProfiler cppipe project files and the pixel-classification software Ilastik. The input to the pipeline can be in either mcd, tiff or txt file format from which stacks of tiff files are generated for subsequent analysis. immunofluorescence/immunohistochemistry data).
This pipeline was generated for Imaging Mass Cytometry experiments, however, it is flexible enough to be applicable to other types of imaging data (e.g. This is an automated image analysis pipeline that sequentially pre-processes and single cell segments imaging data to extract single cell expression data. See main README.md for a condensed overview of the steps in the pipeline, and the bioinformatics tools used at each step.
0 kommentar(er)
