QuPath Repositories

The group’s QuPath-adjacent repos — extensions, scripts, annotation tooling, and PACS integration

← Home · Onboarding

What this is

A curated index of every Memorial repo that touches QuPath. Use this to find the right extension or script before starting a new one.

PACS and scanner integration

Annotation & curation

  • QuickAnnotator — browser-based rapid-annotation interface feeding GeoJSON into QuPath.
  • HistomicsML — interactive region labeling used alongside QuPath.
  • CohortFinder — batch-level QuPath project curation.
  • TissueSelector — tissue-region filtering before annotation.

QC pipelines

WSI viewers and I/O

Downstream DL toolchains

Conventions

  • GeoJSON is the interchange format between QuPath and every downstream tool. Name annotations as {organ}__{class}__{reviewer}.geojson.
  • Scripts live in the project repo where they are used. Only truly generic scripts belong in the extension repos.
  • Don’t commit .qpproj project files for cohorts containing patient identifiers — keep them on the Memorial share.

How to contribute

  1. Pick the repo matching your task (PACS, annotation, QC, viewer, DL).
  2. Read its README; run the bundled example.
  3. Open an issue before building a new script that overlaps with an existing one.

In the research wiki

  • QuPath Tooling — the wiki’s own index of this cluster, kept current from source-level reads rather than from READMEs, with a verdict recorded per repository.
  • QuPath Annotation Workflow — the GeoJSON conventions this page states, written up as a method.
  • WSI Quality Control — what HistoQC, PathProfiler and the GrandQC work each actually cover, and where they do not overlap.
  • Slide Metadata Extraction — reading scanner, magnification, microns-per-pixel and scan date out of the slide file — the prerequisite behind most of the open questions in this cluster.
  • Patch Extraction — the decisions the downstream deep-learning toolchains inherit from whoever cut the tiles.
  • Cross-Stain Registration — the pair-versus-series choice, if slides here ever need aligning across stains.
  • Server-Side Slide Platforms — what a multi-user browser-based platform buys and costs over single-workstation QuPath — the one capability this index does not otherwise provide.
  • Digital Pathology Drafts — the umbrella page for the prototype cluster below — read first, because it records which submodules a fresh clone silently misses.
  • stepbystep — the numbered curriculum and the largest of the prototypes; the first place to look for a working example of a QuPath-to-Python step.
  • Stain Quality — H&E and IHC measurement that independently arrives at the published UK NEQAS method — the closest thing here to a stain-QC metric with outside support.
  • stain-normalisation-qupath — a TIAToolbox wrapper, and the practical route to trying normalisation inside QuPath rather than in a separate pipeline.
  • GrandQC-QuPath — the group’s own wrapper around the published GrandQC artefact models.
  • Outlier Detection via Path Foundation — quality control asked the other way round — embed every tile and score it for distance from the rest, so unusual tiles surface without anyone naming the artefact first.
  • spider-qupath — SPIDER classifiers applied zero-shot, and the cheapest way to see what a general classifier does on local tissue.
  • kaiko-qupath — training a small head on Midnight embeddings — the pattern for turning any cached encoder into a QuPath classifier.
  • metadata-qupath — the Groovy route to scanner metadata, and the worked example behind Slide Metadata Extraction.
  • anonym-qupath — the anonymisation prototype; read De-identification beside it before pointing anything at clinical material.
  • svs-viewer-annotator — three lightweight viewers, useful when QuPath is more than the task needs.
  • Drafts Helpers — OpenSlide fixes and the raw Aperio metadata route, which is where the scan-date parsing lives.