![_images/Tardis_logo_2.png](_images/Tardis_logo_2.png)
Python-based software for generalized object instance segmentation from (cryo-)electron microscopy micrographs/tomograms. The software package is built on a general workflow where predicted semantic segmentation is used for instance segmentation of 2D/3D images.
![_images/workflow.png](_images/workflow.png)
Features
- Robust and high-throughput semantic/instance segmentation of all microtubules:
Supported file formats: [.tif, .mrc, .rec, .am]
Supported modality: [ET, Cryo-ET]
Supported Å resolution: [any best results in 1-40 Å range]
2D micrograph modality microtubule segmentation will come soon!
- Robust and high-throughput semantic/instance segmentation of membranes:
Supported file formats: [.tif, .mrc, .rec, .am]
Supported modality: [EM, ET, Cryo-EM, Cryo-ET]
Supported Å resolution: [all]
High-throughput semantic/instance segmentation of actin [Beta]
Fully automatic segmentation solution!
Napari plugin [Coming soon]
Cloud computing [Coming soon]
Citation
DOI [Microscopy and Microanalysis]
Kiewisz R., Fabig G., Müller-Reichert T. Bepler T. 2023. Automated Segmentation of 3D Cytoskeletal Filaments from Electron Micrographs with TARDIS. Microscopy and Microanalysis 29(Supplement_1):970-972.
Link: NeurIPS 2022 MLSB Workshop
Kiewisz R., Bepler T. 2022. Membrane and microtubule rapid instance segmentation with dimensionless instance segmentation by learning graph representations of point clouds. Neurips 2022 - Machine Learning for Structural Biology Workshop.
What’s new?
- TARDIS-em v0.2.5 (2024-05-22):
Added actin segmentation
Improvement from Microtubule and Membrane prediction with updated models
Added option for scripting TARDIS predictions
Added visualization for semantic and instance predictions
TARDIS build in results visualization
Bug fixes
Documentation tutorials
Pypi and Conda releases
Re-trained DIST model using simulated datasets
- Build 2 model for:
filaments and general 2D structures
3D objects like membranes mitochondria LiDAR data etc.
Quick Start
For more examples and advanced usage please find more details in our Documentation
Install TARDIS-em:
pip install tardis-em
or
conda install tardis-em -c rrobert92
Verifies installation:
tardis
Filaments Prediction
3D Actin prediction
Full tutorial: 3D Actin Prediction
Usage:
recommended usage: tardis_actin [-dir path/to/folder/with/input/tomogram]
advance usage: tardis_actin [-dir str] [-out str] [-ps int] [-ct float] [-dt float]
[-pv int] [-px float] ...
2D Microtubule prediction
TBD
3D Microtubule prediction
Full tutorial: 3D Microtubules Prediction
Example:
![_images/3d_mt.jpg](_images/3d_mt.jpg)
Data source: Dr. Gunar Fabig and Prof. Dr. Thomas Müller-Reichert, TU Dresden
Usage:
recommended usage: tardis_mt [-dir path/to/folder/with/input/tomogram]
advance usage: tardis_mt [-dir str] [-out str] [-ps int] [-ct float] [-dt float]
[-pv int] [-px float] ...
Membrane Prediction
2D prediction
Full tutorial: 2D Membrane Prediction
Example:
![_images/2d_mem.jpg](_images/2d_mem.jpg)
Data source: Dr. Victor Kostyuchenko and Prof. Dr. Shee-Mei Lok, DUKE-NUS Medical School Singapore
Usage:
recommended usage: tardis_mem2d [-dir path/to/folder/with/input/tomogram] -out mrc_csv
advance usage: tardis_mem [-dir str] [-out str] [-ps int] ...
3D prediction
Full tutorial: 3D Membrane Prediction
Example:
![_images/3d_mem.jpg](_images/3d_mem.jpg)
Data source: EMPIRE-10236, DOI: 10.1038/s41586-019-1089-3
Usage:
recommended usage: tardis_mem [-dir path/to/folder/with/input/tomogram] -out mrc_csv
advance usage: tardis_mem [-dir str] [-out str] [-ps int] ...