Skip to content

Latest commit

 

History

History
216 lines (142 loc) · 9.64 KB

index.md

File metadata and controls

216 lines (142 loc) · 9.64 KB
format
revealjs
theme logo slide-number footer transition highlight-style chalkboard controls-layout controls
default
images/JAX_logo_rgb_transparentback.png
true
Imaging Applications, Research IT
slide
arrow
buttons
bottom-right
true

Introduction to CellProfiler

Peter Sobolewski (he/him)

Systems Analyst, Imaging Applications

Research IT

CellProfiler logo https://cellprofiler.org/home

Plan

What is CellProfiler?

CellProfiler is an open-source image analysis program


Key feature:
CellProfiler has a graphical user interface (GUI) for assembling image processing and analysis workflows (called pipelines) using a sequence of modular analysis modules

A bit of background

Strengths of CellProfiler

  • It was designed for batch image analysis using reproducible workflows—without scripting

  • It includes advanced algorithms for image (pre)processing, segmentation, etc. as modular modules

  • It includes in-app documentation for module parameters plus extensive help menu

  • End-to-end workflows: images go in, data tables go out

Limitations

  • Limited viewing capabilities:
    • single planes, single channels only
  • No annotation tools
  • Limited out-of-memory performance
    • lacking support for multiscale ("pyramidal") images
    • lacking built-in tiling/chunking
  • Limited OMERO support
  • Problematic OME.TIFF handling

The CellProfiler concept

  • Create a project that has an associated image file list of locations and an associated pipeline
  • Use the GUI to add a sequence of modules to implement your analysis—these form a pipeline
    • Start with input images and end with measurements in a spreadsheet (.csv) or a MySQL database
  • run in batches over large number of images
  • analyze measurement data in the downstream tool of your choice (Python, R, Excel, etc.)

CellProfiler projects & pipelines

  • projects contain a list of the locations of your image files plus an associated CellProfiler analysis pipeline
    • projects will also contain some cached computed results

  • pipelines can be exported, but do not contain image locations—this is ideal for sharing or reuseing workflows

Typical pipeline

  • start with setting up Input modules
  • Add Image Processing modules to facillitate segmentation
  • Add Object Processing modules to segment the image (identify objects) and manipulate objects
  • Add Measurement modules to get quantified information you're interested in
  • Finish with Data Tools and File Processing to export data or images

CellProfiler UI

{fig-alt="Screenshot of empty CellProfiler UI with menu bar"}

{fig-alt="Screenshot of empty CellProfiler UI with menu bar. At the top right, the Notes box is emphasized with red text. Below, in the middle-right, the Image list drag-and-drop is emphasized with red text."}

{fig-alt="Screenshot of empty CellProfiler UI with menu bar. In the top left, the four Input modules are highlighted with a red box and red text."}

{fig-alt="Screenshot of empty CellProfiler UI with menu bar. On the middle-left, the area for adding pipeline modules is highlighted with red text. Below, the module buttons are highlighted with a red box."}

{fig-alt="Screenshot of CellProfiler UI with menu bar and a single module (ColorToGrey) open. At the top, the Test menu is open and the Start Test Mode option is highlighted with a red box. At the bottom left, the Start Test Mode button is highlighted with a red box."}

{fig-alt="Screenshot of CellProfiler UI with menu bar and a single module (ColorToGrey) open. Test mode is now active. At the top of the modules area, the Test mode buttons next to the ColorToGrey module are highlighted with a red box. At the bottom left, the Run and Step buttons are highlighted with a red box."}

{fig-alt="Screenshot of CellProfiler UI with menu bar and a single module (ColorToGrey) open. Test mode is now active. At the top, the Test menu is open and the Image set menu items are highlighted with a red box. At the bottom left, the Next Image Set button is highlighted with a red box."}

{fig-alt="Screenshot of CellProfiler UI with menu bar and a single module (ColorToGrey) open. In the middle right, the ColorToGrey module options are showing with the input image selection and output image naming highlighted with red boxes."}

Keys to success

  • ensure image inputs are properly set up (NamesAndTypes) and that they and intermediate step outputs have meaningful names
  • use Test mode to see the outputs of individual modulues (pipeline steps) one-at-a-time
    • let's you test parameters on a single image or any number of images (even a randomly selected one)
    • can run the whole pipeline or just individual modules

Getting more help

Segmentation Tutorial

  • Ensure you have the tutorial materials downloaded. If not, download the .zip file
  • Inside the folder is a PDF walkthrough, if you want to follow along on your own

Import the starting pipeline

  • Inside the folder are also two .cppipe files, which define the CellProfiler pipelines
    • segmentation_start.cppipe is our starting point, it handles the initial image importing and set up.
    • Drag-and-drop it onto the Pipeline Modules section (middle left pane) to import it.

:::{.callout-tip} The segmentation_final.cppipe is the final pipeline, which you can use to see the completed analysis pipeline! :::

Import the images

  • Import the images by drag-and-dropping the images_Illum-corrected folder onto the central image list pane
  • Click on Apply filters to the file list to filter out non-images

:::{.callout-tip} Double-click an image in the list to open and view it :::

Completed setup

{fig-alt="Screenshot of CellProfiler UI with menu bar and the segmentation_start pipeline imported. The image list is populated with images and an example is open."}

Input Modules Tutorial

This tutorial will introduce you to Input Modules in CellProfiler: Images, Metadata, NamesAndTypes, and Groups. These modules are crucial for loading and organizing your input data so CellProfiler manages it properly in the analysis pipeline.

:::{.callout-important} Setting up Input Modules properly is crucial to being able to analyze your data! :::

Input Modules

  • Images module: set where your images are located, apply filters to include/exclude certain files
  • Metadata module [optional]: extract and associate metadata with your images, e.g. well plate layout, treatment, time points
  • NamesAndTypes module: give each image a meaningful name by which modules in the analysis pipeline will refer to it
  • Groups module [optional]: group images together to be analyzed independently, typically by field-of-view, e.g. z-stack, time course

Getting started

:::{.callout-tip} Each example has .cppipe file, which has the "solution" :::

Metadata regex tips

  • The https://regex101.com/ website is great! Just ensure you have one of the PCRE flavors selected.
  • Copy-paste the file name you want to parse into the test string box
  • Capture groups are always of the format:
    (?P<Name>[what]{how many})
    • Name is what you will refer to in the Metadata module
    • what can be digits, e.g. 0-9, or capital letters, e.g. A-Z, etc.
    • how many is an integer, e.g. 1, 2, 3, etc.

A few more regex tips

  • . is a wildcard for any character
  • * is a quantifier for 0 or more consecutive of the preceding character/token, e.g.: [A-Z]* to match any number of capital letters
  • ^ is an anchor for the start of the string
  • $ is an anchor for the end of the string--don't forget about extensions!
  • \ is an escape character, e.g. \. to match a period, rather than any character