View markdown source on GitHub

Nucleoli Segmentation
Feature Extraction
using CellProfiler




last_modification Last modification: Jul 10, 2021

What is the nucleoli?

.image-55[ image of four cells, one is dividing, and red arrows point to the 5 nucleii. ]

Speaker Notes In the DNA channel, the nucleoli is shown as the absence of DNA (red arrows). —


.image-65[ screenshot of a complex tool interface showing an array of images (1) and then two selected images below it (2) and a configuration panel on the right (3). ]

Speaker Notes In the IDR the images can be selected in the user interface (1). The images will then show up at the bottom (2) where they need to be selected once again and then copy the URL from the link icon on the top left (3). That URL contains the image ids that will be downloaded. You can also bring your images by uploading the DNA channel of your images to your Galaxy history. —

General Workflow

screenshot of a galaxy workflow with four sections highlighted by boxes reading nuclei, nucleoli, background, and feature extraction.
Figure 1: High-level view of the workflow

Speaker Notes

1) Segment nuclei

Inset to the first portion of the above overview image showing nuclei with 5 steps, starting modules, segment complete nuclei, segmentation mask complete nuclei, label nuclei, and save labelled nuclei.

Speaker Notes Every CellProfiler pipeline needs to start by processing the metadata with the tool “Starting modules”.

Segmentation and labelling of the nuclei

a photo of cells in black and white, several cells have red numbers written atop them.
Figure 2: Identified nuclei with labels

Speaker Notes Resulting image after the first logical step of the workflow in which the nuclei were segmented and labelled. —

2) Segment nucleoli

Inset of the nucleoli portion of the workflow with the following steps: detect dark holes in nuclei, segment nucleoli that fall inside nuclei, segment nucleoli, convert the segmented nucleoli into an image, combine masks (nuclei+nucleoli), and finally save combined segmentation masks.

Speaker Notes To identify the nucleoli the first step is to enhance the dark holes so that they can be segmented. The segmentation of the dark holes will take place only when the dark holes fall inside a nucleus. This is to avoid the detection of wrong objects outside the cell. At this step, the segmentation of the nucleoli can be performed. It will be useful for visual exploration that the masks of the nuclei and nucleoli are saved together into one segmentation mask. —

Combined mask: nuclei + nucleoli

image of the cells again, but now cells are blue and nuclei/nucleoli are red.
Figure 3: Nuclei and nucleoli masks combined in which the nuclei are in blue and nucleoli in magenta.

Speaker Notes The combination of both masks gives an image with the nuclei in blue and the nucleoli in magenta. —

Background extraction

third inset workflow, background extraction, with steps segment all nuclei, segmentation mask nucleoli including cells touching borders, and extract background.

Speaker Notes

The extraction of the background can be useful to measure features that indicate that the images are artefacts. To get the background we first select the foreground and subtract it from the original image. We have partially detected the foreground in the nuclei segmentation in previous steps. However, that mask only included just those objects within a max and a min pixel size. We also discarded the images touching the borders. Here, we get rid of the constraints and segment all the nuclei.

Feature extraction

The final inset, feature extraction with very small steps.

Speaker Notes The features can be extracted from the images and the objects. Depending on which type of object and the interest of the study, different parameters can be measured. The module that relates the objects is useful to export a table with the ids of the nucleoli in relation with their parent nuclei. That’s useful for a meaningful interpretation of the data.

Key Points

Thank you!

This material is the result of a collaborative work. Thanks to the Galaxy Training Network and all the contributors! Galaxy Training Network This material is licensed under the Creative Commons Attribution 4.0 International License.