MedicoSAM implements interactive annotation and (automatic) semantic segmentation for medical images. It is built on top of Segment Anything by Meta AI and specializes it for biomedical imaging data. Its core components are:

• The medico_sam publicly available model for interactive data annotation in 2d and 3d data that are fine-tuned on publicly available medical images.
• The medico_sam library provides training frameworks, inspired by Segment Anything for Microscopy, for downstream tasks:
  • Apply Segment Anything to 2d and 3d data or fine-tune it on your data.
  • Supports semantic segmentation for 2d and 3d data.

Based on these components, medico_sam enables fast interactive and automatic annotation for medical images.

How to install medico-sam python library from source?

To create the environment and install medico_sam into it follow these steps:

1. Clone the repository: git clone https://github.com/computational-cell-analytics/medico-sam
2. Enter it: cd medico-sam
3. Create the environment with the necessary requirements: conda env create -f environment.yaml
4. Activate the environment: conda activate medico-sam
5. Install medico_sam: pip install -e .

You can find the model checkpoints at: https://owncloud.gwdg.de/index.php/s/AB69HGhj8wuozXQ

Download it via terminal using: wget https://owncloud.gwdg.de/index.php/s/AB69HGhj8wuozXQ/download -O vit_b_medicosam.pt.

See TOOL_USAGE.md document for details.

TLDR: We recommend using our model with micro-sam annotator tool, in terms of compatibility and ease of annotation experience!

If you are using this repository in your research please cite:

• our preprint.
• the Segment Anything fo Microscopy publication.
• and the original Segment Anything publication.