Method / tool

QuPath

Open-source whole-slide image analysis platform for digital pathology; combines interactive annotation, scriptable batch workflows, and a tissue-aware object model in a single extensible environment.

The problem — Whole-slide images (WSIs) are gigapixel files. Standard bioimage tools built for single-field fluorescence data cannot open them, let alone run reproducible batch analysis across a cohort. Without a structured analysis environment, pathology workflows collapse into manual annotation on proprietary viewers, with no audit trail and no reuse path.

What it is / how it works — QuPath is a Java-based open-source platform purpose-built for WSI analysis. It models tissue as a hierarchy of objects — annotations, detections, and cells — rather than a flat pixel array. Users define regions interactively or via script, then apply detection algorithms (nucleus detection, cell expansion, positive-cell scoring) and classifiers across those regions. Groovy scripting exposes the full object model, enabling reproducible, version-controlled pipelines. QuPath supports every major WSI format via OpenSlide and its own Bio-Formats bridge. The original paper (Bankhead et al. 2017, Scientific Reports) established its core architecture; version 0.7.0 ships today under the GPL.

The platform's tissue-aware object model makes it a natural fit for the segment and features stages of a digital-pathology pipeline, and its annotation exports wire cleanly into downstream specialist tools for spatial statistics or machine-learning readout.

Where it breaks — QuPath's interactive paradigm makes reproducibility easy to undermine. Classifiers trained by one analyst on one slide set do not transfer automatically; color normalization is platform-default unless overridden; and the Groovy script environment has no dependency management, so a pipeline that runs today may not run against a different QuPath version. The tissue hierarchy is also a trap: detection parameters chosen at 20× look wrong at 40×, silently inflating or deflating cell counts without error.

A QuPath project is not a pipeline. Scripts are reproducible only if they are version-controlled alongside the QuPath release they target, the color normalization settings are explicit, and the classifier training data is committed. Without those three constraints, [off-the-shelf reproducibility](/lab-notes/choosing-a-bioimage-analysis-pipeline-off-the-shelf-vs-custom) is an illusion even when the tool is free and open-source.

References

Appears in these notes

  • Choosing a Bioimage Analysis Pipeline — Off-the-Shelf vs CustomMost microscopy questions are answered by composing existing, validated tools — and a custom pipeline is justified only when off-the-shelf options fail on a specific, identifiable axis. This note gives the decision criteria and the signals that you have genuinely crossed into custom territory.
  • Why Interoperability Matters in MicroscopyA microscopy pipeline is always several tools in sequence, so the seams between them — file formats, metadata, coordinate conventions — decide whether the pipeline composes or quietly corrupts. Open standards like OME-Zarr/NGFF turn hand-offs from lossy conversions into clean interfaces, and avoid lock-in to any one vendor or tool.
  • The Bioimage Analysis Ecosystem — Where Each Tool FitsBioimage analysis is not one tool but a landscape organized along two axes — interactive vs scriptable, and generalist vs specialist. Fiji, napari, CellProfiler, QuPath, ilastik, and the deep-learning segmenters each occupy a different cell of that grid, and choosing well means knowing which cell your problem lives in.

← Back to the constellation