Local & self-hosted appsQuPath open-source project (University of Edinburgh)Open source
QuPath
Decision matrix
QuPath app decision matrix
Overview
QuPath is free, open-source software for digital pathology, with built-in cell detection, interactive machine-learning classifiers and whole-slide image support. Pick it when you need to quantify what's on a tissue slide. It is explicitly not for clinical or diagnostic use, so keep it to research analysis.
Other good fits
Training a quick interactive classifier by clicking examples on the slide instead of hand-coding rules.
Scripting a repeatable analysis in Groovy once you've confirmed a method works on a few slides.
+ Pros
Read pros
QuPath is open-source, GPLv3-licensed software built specifically for digital pathology and bioimage analysis. It reads whole-slide images through Bio-Formats and OpenSlide, so scanner output that chokes a generic image viewer opens normally. Its built-in algorithms handle common tasks out of the box: cell and tissue detection, and interactive machine-learning classifiers for both objects and pixels that you train by clicking examples directly on the image rather than writing detection rules from scratch.
Pick it when
You have whole-slide scans and need to detect, count or classify cells or tissue regions.
You want a classifier trained interactively on your own slide, not a black-box pretrained model.
You need the analysis to be repeatable across many slides via Groovy scripting once it works on a few.
− Cons
Read cons
QuPath explicitly states it is "not intended for clinical, diagnostic or therapeutic purposes," so any result feeding a clinical decision needs a validated, regulated tool instead. It is a desktop application with no hosted version, and while scripting scales an analysis across many slides, building and validating that script still takes real time per project. For anything short of whole-slide pathology work, such as a single fluorescence image, its interface carries more machinery than the job needs.
Pick something else when
The result needs to inform a clinical or diagnostic decision: QuPath's own documentation rules that out; use a validated, regulated tool.
You're viewing a single multi-channel microscopy stack rather than a whole-slide scan: napari is a better fit.
You need cell segmentation as a standalone step to feed into another pipeline: Cellpose is the more focused tool, and its output can be brought into QuPath if needed.
Avoid if
You only need to view or lightly edit a normal microscopy image, not a whole-slide scan; Fiji or napari is a lighter fit.
You need a scripted sequence- or structure-analysis pipeline rather than image quantification; use Biopython instead.
Workflow
Open a whole-slide image (or a smaller region) in QuPath → run the built-in cell or tissue detection to get an initial set of objects → train an interactive classifier by clicking correctly and incorrectly classified examples → review the classifier's output across the slide and correct where it's wrong → once satisfied, save the analysis as a Groovy script and run it unattended across the rest of your slide set.
Cost
QuPath is free and open source under GPLv3, with no paid tier, licence fee or usage cap. It installs on Windows, macOS (Intel and Apple silicon) and Linux, with no cloud account required. The running effort is validating that a detection or classifier trained on one slide still performs on the next one, since staining and scanning can vary batch to batch, and keeping scripts up to date as QuPath releases new versions.
Restrictions
QuPath's own documentation states it is "not intended for clinical, diagnostic or therapeutic purposes," restricting it to research use. It is desktop-only software with no official cloud or hosted deployment.
First project
Documentation-based suggestion — not run here.
Read first project
Detect and classify cells in one region of a slide, then script the same analysis across the full set. The book Quantifying a tissue slide with QuPath walks through this.
Open a whole-slide image and select a representative region to work in first.
Run built-in cell detection and inspect how well it finds real cells versus artifacts.
Train an interactive classifier by clicking a handful of correct and incorrect examples.
Apply the classifier to the whole region and visually check the result against the raw image.
Export the workflow as a Groovy script and run it on two more slides to confirm it generalises.
Links & evidence
Documentation-based · not hands-on · Sep 23, 2026
Rests on the QuPath project page and its ReadTheDocs documentation (about and licence pages), all read on 2026-09-23, for capabilities, platforms and licensing. Nothing was installed or run for this review; the first project is a suggested exercise.
QuPath project page · Sep 23, 2026 — Open software for bioimage analysis; current stable version 0.7.0; documentation and image.sc community forum linked.
QuPath documentation: about · Sep 23, 2026 — Built-in cell/tissue detection, interactive object and pixel classifiers, whole-slide support via Bio-Formats/OpenSlide, Groovy scripting; explicitly not for clinical, diagnostic or therapeutic use.
QuPath license and acknowledgements · Sep 23, 2026 — QuPath is open-source software available under the GNU General Public License v3.0.
Further review notes & alternatives
Alternatives and trade-offs
Fiji is the better fit for general-purpose image processing on single images or small stacks, rather than whole-slide pathology.
napari wins when the data is a multi-dimensional microscopy stack you want to view and annotate in Python, not a pathology slide.
Cellpose is a narrower, more accurate option specifically for cell/nucleus segmentation, which you can feed into QuPath rather than relying solely on QuPath's own detector.
QuPath is open-source software for digital pathology and bioimage analysis, with built-in cell and tissue detection algorithms, interactive machine-learning classifiers, and whole-slide image support via Bio-Formats and OpenSlide. It runs on Windows, macOS and Linux and scripts in Groovy, but it is explicitly not for clinical, diagnostic or therapeutic use.
QuPath open-source project (University of Edinburgh) · free · desktop