pyMarAI: nnU-Net-based Tumor Spheroids Auto Delineation

pyMarAI is a toolchain including a PyQt5-based graphical user interface that allows to apply a CNN-based delineation workflow for bioimaging-driven tumor spheroid growth assays commonly used in cancer research.

The software provides a complete pipeline for handling microscopic spheroid image data, generating deep-learning–based 2D delineations, and allows to curate results for continuous model improvement.

[!IMPORTANT] Regulatory status: This software and the bundled model are intended solely for research and development (R&D). They are not intended for primary diagnosis, therapy, or any other clinical decision-making and must not be used as a medical device.

Features

• User-friendly GUI (PyQt5)

  • Intuitive interface for configuring and running CNN-based delineation workflows without having to work directly on the command line.

• Image data management

  • Import and organize spheroid image datasets
  • Automatic conversion between supported image formats (PNG, TIFF)
  • Management of large-scale spheroid experiments (>10000 of spheroid images)

• CNN-based delineation at scale

  • Ships with a pre-trained network model trained on microscopic spheroid images
  • Run automated delineation on large spheroid datasets
  • Support for GPU-accelerated environments to speed up inference

• Quality review and curation

  • Visual review of delineated images
  • Tagging of delineation quality (GOOD, BAD)
  • Optional manual correction hand-off to external applications

• Continuous dataset growth for retraining

  • All new and manually corrected delineations are collected in a centralized directory structure
  • Facilitates expansion of the training dataset for future retraining and further improvement of the CNN model

Supported formats

pyMarAI allows to process the following microscopic image data formats:

• PNG / TIFF microscopic images

Requirements

• Python 3.10+
• nnUNet v2
• optional: dedicated conda environment

Build & install

pip install .

Model installation

pyMarAI works with nnUNet v2+ trained network models and provides the possibility to define the nnUNet-based network model which will be used for interference in a dedicated configuration file. In addition, it ships with a pre-trained network model which has been trained on thousands of microscopic spheroid images (see 'Citation' for the corresponding publication).

To install and use this pre-trained network model please refer to the following data publication where you can download the network model from:

https://doi.org/10.14278/rodare.4198

Citation / Publication

If you use pyMarAI (or parts of it) in your own projects, evaluations or publications please cite our corresponding publication using:

@article{Maus2026,
  title = {Automatic Delineation of Tumor Spheroids in Microscopic Images Using Deep-Learning},
  author = {Maus, Jens and Nitschke, Janina and Nikulin, Pavel and Hofheinz, Frank and Barth, Mareike and Lemm, Sandy and Richter, Lena and Pietzsch, Jens and Braune, Anja and Ullrich, Martin},
  journal = {ACS Measurement Science Au},
  publisher={ACS Publications},
  year={2026},
  doi = {10.1021/acsmeasuresciau.5c00172}
}

The publication is licensed under CC-BY-4.0.

Disclaimer (Research Use Only – Not a Medical Device)

This software and any bundled or referenced model weights are provided exclusively for research and development purposes. They are not intended for use in the diagnosis, cure, mitigation, treatment, or prevention of disease, or for any other clinical decision-making.

• The software is not a medical device and is not CE-marked.
• No clinical performance, safety, or effectiveness is claimed or implied.
• Any results must not be used to guide patient management.
• Users are responsible for compliance with all applicable laws, regulations, and data protection requirements when processing data.

THE SOFTWARE AND MODELS ARE PROVIDED “AS IS”, WITHOUT ANY WARRANTY, EXPRESS OR IMPLIED.

Licenses

The code in this repository is licensed under Apache-2.0 (see LICENSE).
The model weights are licensed under CC-BY-SA-4.0 (see MODEL_LICENSE.md).

Third-Party Licenses

This project uses or interoperates with the following third-party components:

• nnUNet v2 – Copyright © respective authors.
  • License: Apache-2.0
• PyTorch, NumPy, Nibabel, etc.
  • Licensed under their respective open-source licenses.

Each third-party component is the property of its respective owners and is provided under its own license terms. Copies of these licenses are available from the upstream projects.

Notes on Qt licensing

pyMarAI itself is released under Apache-2.0. It uses pyQt, which is available under GPLv3. If you distribute applications that use pyMarAI you are responsible for GPL compliance for pyQt (dynamic linking recommended, include license texts, do not prohibit relinking, and provide installation information for locked-down devices).