Leishmania Morphology Image Analysis Pipeline

This pipeline works on three-channel microscopy images of fixed Leishmania mexicana cells - a phase contrast image, a DAPI channel and an mNeonGreen gene channel. The segmentation is performed on the phase image, with some input from the DAPI channel (the location of the kinetoplast).

Quickstart

The component doing the image segmentation and analysis is the ImageSet class, which takes a path to a directory containing the three channel tif files. It's identified_cells property then contains a SegmentedCellCollection with a list of SegmentedCells. A SegmentedCell object holds information on a single observed cell, such as length, width, flagellum length and a thumbnail image.

import leishmorphology

im = leishmorphology.ImageSet(path_to_imageset)
cells = im.identified_cells

To analyse a collection of image sets, one can use the SegmentedCellCollectionSet class, which also allows saving of the produced data in a HDF5 file (via pandas and pytables):

import leishmorphology

cellset = leishmorphology.SegmentedCellCollectionSet([
    leishmorphology.segmentation.ImageSet(path).identified_cells for path in [path1, path2, path3])
cellset.to_hdf("cells.h5")

Loading of a collection works via the static method SegmentedCellCollectionSet.load_hdf:

import leishmorphology

cellset = leishmorphology.SegmentedCellCollectionSet.load_hdf(path_to_hdf5_file)

Displaying data

The SegmentedCell, SegmentedCellCollection and SegmentedCellCollectionSet classes have rich display methods which integrate into IPython's/Jupyter's rich display infrastructure. Therefore the following happens:

• A SegmentedCell instance displays its thumbnail with the segmented cell and (if found) its flagellum highlighted. It also displays a box with cell measurements.
• A SegmentedCellCollection instance displays a floating grid of the SegmentedCells it contains.
• A SegmentedCellCollectionSet instance displays a scatter plot of cell width vs length (with markers distinguishing cells from different SegmentedCellCollections) and colour showing the flagellum length (if present).

SegmentedCell is a python dataclass with each attribute describing a property of the segmented cell. SegmentedCellCollection has a name attribute which, if not given, defaults to the image set path. It contains a cell_list attribute which simply holds a list of SegmentedCells. Finally, SegmentedCellCollectionSet has a single attribute sets which holds a list of SegmentedCellCollections. The latter also has a method to_dataframe which returns a pandas.DataFrame table with one cell per row.