Releases: Zuber-group/CryoVesNet
v0.5.1
v0.5.1 Release
CryoVesNet v0.5.1 - Minor Update
This update introduces a Pyto wrapper for segmentation, enhances compatibility, and updates documentation.
🎉 Updates
- Pyto Wrapper: Now compatible with Pyto v1.9.2 for smoother integration.
- Documentation: Updated README with the latest DOI and new video resources.
🔬 Research Application
CryoVesNet is published in Journal of Cell Biology. Access the article at https://doi.org/10.1083/jcb.202402169.
v0.5.0
CryoVesNet v0.5.0 - Public Beta Release
We're excited to announce the first public beta release of CryoVesNet! This release marks a significant refactor as we open our deep learning-based tool for automatic segmentation of synaptic vesicles in cryo-electron tomography (cryoET) data to the wider scientific community for testing and feedback.
🎯 Key Features
- Automatic segmentation of synaptic vesicles in cryo-ET data
- Generalization across different datasets (rat synaptosomes and primary neuronal cultures)
- Applicability to any spherical membrane-bound organelle
- Pre-trained network and custom training capabilities
🚀 Major Updates
Enhanced Deep Learning Pipeline
- Multi-GPU support for faster processing
- 2D Network support with new architectures (ResNet, Inception, VGG)
- Adaptive vesicle labeling for improved accuracy
- Included elliptical vesicles for improved vesicle shape analysis
Improved Visualization and Interaction
- Interactive label editor for easy modification of segmentation results
- Undo functionality for manual corrections
- Real-time feedback with immediate visual updates
Performance Optimizations
- Optimized collision solver for overlapping vesicles
- Batch processing for improved speed on large datasets
Enhanced Initialization and Memory Efficiency
- Added
patternparameter to__init__for more flexible file matching - Optimized memory usage by changing various int types to int16
- Improved overall memory management across pipeline steps
Project Structure and Dependencies
- Updated package information (name: 'CryoVesNet', version: 0.5.0)
- Expanded compatibility with latest numpy, scipy, and scikit-image versions
- Python 3.9+ requirement
🔧 New Additions
- OS Compatibility: Tested on Ubuntu 20.04.3 LTS, macOS and Windows 11
- Interactive cleaning tool for manual refinement of segmentation results
- Flexible pipeline options for customizing the segmentation process
- GPU support instructions for Windows users
📊 Performance
- Runtime: ~200 seconds for a typical tomogram on a single GPU (A4000 Nvidia GPU)
- ~27.5 minutes on a Macbook Pro M1 without GPU utilization
📘 Documentation
Comprehensive documentation, including detailed installation instructions, usage examples, and folder structure, is available in the README.md file of the repository.
🔬 Research Application
If you use CryoVesNet in your research, please cite:
https://doi.org/10.1101/2024.02.26.582080
We welcome your feedback and contributions to CryoVesNet. Thank you for your interest in our project!