MEDiml-app is a desktop application for extracting radiomics and dosiomics features from medical images without writing code. It provides a graphical user interface to the MEDiml Python package and is part of the MEDomics project.
Radiomics turns medical images (CT, MR, PET, radiotherapy dose maps) into quantitative features that can feed predictive models of diagnosis, treatment response or outcome. Computing these features reproducibly is hard: the results depend on many preprocessing choices (voxel size, re-segmentation range, discretisation, filters), and the standardized tools that implement them, such as MEDiml, which follows the Image Biomarker Standardisation Initiative (IBSI), are programming libraries.
Many of the people who need radiomics features, such as clinicians, medical physicists and clinical researchers, do not program. MEDiml-app closes that gap. It lets them:
- convert DICOM or NIfTI datasets into MEDiml's format and check them before extraction;
- start from ready-to-use extraction settings, with in-app explanations of what each setting changes in the features;
- extract radiomics and dosiomics features for a whole dataset in one run;
- keep, with every result, the exact settings that produced it, so studies can be reported and repeated.
It also gives researchers who do program a fast way to prototype MEDiml settings before scripting them.
- Graphical user interface: every MEDiml step, with no code to write.
- Data processing and pre-checks: DICOM and NIfTI conversion, plus voxel and intensity range checks across a dataset.
- Batch extraction: radiomics and dosiomics features for a whole dataset, with editable, validated settings.
- Data analysis: machine learning models for the extracted features.
- Customizable workflows: design and run workflows tailored to your study.
Download the installer for your operating system from the releases page and follow the Quick start guide.
| Operating system | Installer |
|---|---|
| Windows 10 or 11 (64-bit) | .exe |
| Ubuntu 22.04 or later (64-bit) | .deb |
| macOS (Apple silicon) | .dmg |
- A 64-bit operating system from the list above.
- An internet connection on first launch: the app downloads its own Python 3.9 environment (with MEDiml) into
~/.mediml/python, and MongoDB. No Python installation is needed beforehand.
Prerequisites: Git, Node.js 18 and Go 1.21 or later (Go 1.24 or later on macOS).
git clone https://github.com/MEDomicsLab/MEDiml-app.git
cd MEDiml-app
# 1. JavaScript dependencies
npm install
# 2. Go server (rebuild it after any change to go_server/)
cd go_server
go build main.go
cd ..
# 3. Start the app in development mode
npm run devPython environment: on first launch, the setup dialog installs Python 3.9 and the packages listed in pythonEnv/requirements.txt (MEDiml included) into a python/ folder in the repository, as well as MongoDB. If ~/.mediml/python already exists, for example from an installed copy of the app, it is used instead.
To build the installers yourself, run npm run build:win, npm run build:linux or npm run build:mac on the matching operating system. The files are written to build/dist/.
New to MEDiml-app? Follow the Soft-Tissue Sarcoma proof of concept, an easy-to-follow guide to get familiar with the app. It walks through a full radiomics study for the early evaluation of lung metastasis risk, from feature extraction to model training and evaluation.
Its imaging data (PET, T1, T2FS and CT scans for each patient, derived from a public TCIA cohort) is published on Zenodo. You can download it from the app's home page, straight into your workspace under DATA/STS_POC.
- MEDiml-app documentation, with the Quick start and tutorials
- MEDiml Python package and its documentation
- Bugs and feature requests: open an issue on GitHub Issues.
- Questions and discussion: join the MEDomics Discord.
Contributions are welcome. See CONTRIBUTING.md and our Code of Conduct.
If you use MEDiml in your research, please cite it using:
@misc{loutfi2026mediml,
title = {Medical Image Analysis with Machine Learning ({MEDiml}): Strategies for Simpler Radiomics},
author = {Ait Lhaj Loutfi, Mahdi and others},
year = {2026},
eprint = {2610.37715},
archiveprefix = {arXiv},
doi = {10.48550/arXiv.2610.37715}
}The full reference manual of MEDiml is available on arXiv.
MEDiml-app is released under the GNU General Public License v3.0.