Introduction to the BBBC021 dataset

  • BBBC021 is a high-content imaging dataset of MCF7 cells.
  • File names contain information about treatments and experimental parameters.

Reading imagesConclusions

  • Load images by dragging/dropping them onto CellProfiler’s Images module
  • The Metadata module translates file names to extract metadata from file names, which will be saved along with your measurements.
  • The NamesAndTypes module converts image names to meaningful names to be used within CellProfiler.

Identifying primary objects

  • Nuclei are often the best primary objects because they have high contrast and exist in nearly all cells.
  • The most important parameters in IdentifyPrimaryObjects are object size, thresholding method, and declumping.
  • Always test segmentation on multiple images and across experimental conditions.

Identifying secondary & tertiary objects

  • Secondary objects (cells) are typically grown from primary objects (nuclei) using a cytoplasmic/cell-boundary stain (here: actin).
  • Filtering border-touching nuclei helps avoid partial cells and misleading measurements.
  • The most important settings in IdentifySecondaryObjects are the identification method and thresholding choices, which strongly affect whether cells merge or fragment.
  • Tertiary objects (cytoplasm) are a subtraction of nuclei from the cell mask.

Measuring object intensity and shape

  • MeasureObjectIntensity quantifies fluorescence per object; choose objects and channels deliberately.
  • MeasureObjectSizeShape quantifies morphology; disable Zernike/advanced features to iterate faster.

Reproducibility in CellProfiler

  • Save Project is great for restoring a full session, but often stores image paths that are not portable.
  • Export/Import Pipeline is a more reproducible way to share an analysis workflow.

Bonus: visualising features

  • CellProfiler writes one .csv per object type (i.e. Nuclei, Cells, Cytoplasm).
  • Exported files contain many columns with metadata.
  • Morpheus can be useful to interrogate morphological changes.

Advanced: classifying cells in CellProfiler Analyst

  • CellProfiler Analyst can classify cellular phenotypes using the measurements exported by CellProfiler.
  • The properties file tells CPA how to find and interpret your database and images.
  • ExportToDatabase can generate CPA-ready outputs, but database may need extra filtering.
  • Classifier performance depends heavily on consistent labels and representative training examples.