The programmed death of manual lab pipelines.

Drop in raw microscope files, let AI find every cell, review the masks yourself, and compare conditions — in the browser, on your own computer. Open source.

Install dependencies

apoptotic-lab segmenting a fluorescence field: 178 nuclei found, each outlined over the raw image

Validated against a published StarDist workflow — 5 real datasets, 78 fields, 21,385 reference nuclei

0.9994Lin’s concordance, nuclei per field

0.948–0.964Dice mask overlap, every dataset

±1.2 %per-cell intensity bias, every marker

−1.5 %mean difference in counts per field

From raw files to statistics in four steps

Raw data step: channels of a CZI file detected and previewed — **Raw data** Drop in CZI or TIFF files at native bit depth; channels are read and previewed as the microscope saved them.

**Raw data** Drop in CZI or TIFF files at native bit depth; channels are read and previewed as the microscope saved them.

Segment step: nuclei outlined by Cellpose-SAM — **Segment** Cellpose‑SAM or StarDist finds every nucleus or cell on your GPU. Every mask can be corrected by hand before anything is measured.

**Segment** Cellpose‑SAM or StarDist finds every nucleus or cell on your GPU. Every mask can be corrected by hand before anything is measured.

Measure step: per-cell intensity table — **Measure** Every marker, inside every cell: areas, positions and per-channel intensities, exported as CSV.

**Measure** Every marker, inside every cell: areas, positions and per-channel intensities, exported as CSV.

Analysis step: distributions and positivity against the control — **Analysis** Distributions, co‑expression and the percentage of positive cells against your control, with publication‑ready plots.

**Analysis** Distributions, co‑expression and the percentage of positive cells against your control, with publication‑ready plots.

Local by design

The engine runs on your computer and uses its GPU. Patient-derived images never leave the lab, and nothing is uploaded.

You stay in control

The decisions that matter scientifically — which channel, which control, which mask is wrong — stay with the researcher.

Reproducible by default

Every export carries a record of exactly how it was produced: model, settings, thresholds. Results you can defend to a reviewer.

Run it on your computer

The interface is this page. The engine installs once — Python 3.10 to 3.12 and Git, then:

git clone https://github.com/abrilcode/apoptotic-lab.git
cd apoptotic-lab
python3 -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
uvicorn app:app --host 127.0.0.1 --port 8010

Windows steps are in the installation guide. Then:

New analysis

Segment nuclei, cells or other objects on any channel, measure every channel inside each object, and compare conditions. Images are processed on this computer.

Segmentation deviceChecking

StarDist (optional, nuclei)Checking

Output folderNot chosen; downloads go to the browser

DependenciesChecking

Some dependencies are missing or out of date

Install them in the engine's environment, then restart the engine and reload this page.

Installation steps on GitHub

Package	Required	Installed

Raw data

One image: a multichannel file, or the single-channel files of one field.

Drop the image here, or click to choose CZI, TIFF, PNG or JPEG

Measure

Every channel measured inside each object. Each table comes with its parameter record and the final masks.

Set measurements

Measurement options

Subtract backgroundi

Pixel size µm per pixeli

Analysis

Distribution of one measurement per channel. Descriptive only.

From raw files to statistics in four steps

Local by design

You stay in control

Reproducible by default

Run it on your computer

New analysis

Some dependencies are missing or out of date

Raw data

Segment

What to segment

Measure

Set measurements

Measurement options

Analysis

Raw data

Raw data

Conditionsi

Images

View

Start the engine on this computer

From raw files to statistics in four steps

Local by design

You stay in control

Reproducible by default

Run it on your computer

New analysis

Some dependencies are missing or out of date

Raw data

Segment

What to segment

Measure

Set measurements

Measurement options

Analysis

Raw data

Raw data

ConditionsiCondition is the group being compared. Technical replicate is the field number (Foto<n>) inside that condition: another imaged position of the sample, not an independent biological replicate. Click a thumbnail to inspect the channels.

Images

View

Conditionsi