Releases: bowang-lab/MedSAMSlicer
Releases · bowang-lab/MedSAMSlicer
v0.13
Change log:
- A fast and accurate version of MedSAMLite called MedficientSAM is included based on this repository
v0.12
Change log
- Proper dependency check
- Colored image segmentation problem fixed
- New automated tests added
- Various changes introduced to conform with 3D Slicer Extension Index
v0.11
Faster and more diverse segmentation
Change log:
- OpenVINO engine is added for a faster segmentation for users without an NVIDIA GPU.
- DAFT (a faster segmentation model) is added based on this code base.
- Faster segmentation can now be achieved at the cost of a moderate accuracy loss.
- Several issues including #24, #30, #34, #37 are addressed.
v0.1
Major changes has introduced to the plugin to make it ready to be published on 3D Slicer Extension Index Catalog. Among the main changes, we have:
- Client-Server arch for the module has changed. It was the main source of bugs and problems in previous versions
- We now support "Save Scene" functionality of 3D Slicer.
- More flexibility has been introduced to the preprocessing step.
- The plugin is slightly faster in both load time and runtime
v0.06
Changelog:
- Boundary selection on R-S slice is now possible
- Upgrade issue fixed
- Usage instructions updated
v0.05
Changelog:
- Single segmentation feature is added: if you need to segment only a small part of the image, there is no need to "send image", which is a lengthy process
- Boundary selection bug fixed
- Progress-check timer set up fixed
v0.04
Changelog:
- Fine-tuned models can be loaded as well as default pre-trained model
- Inconsistencies in the backend are resolved
- Issue #6 addressed
- Minor GUI improvements
v0.03
Change log:
- In-module upgrade button
- Issue #1 addressed: Non-square slices are processable
- Issue #7 addressed: Including server installation using local zip files
v0.02
Change log:
- Uses GPU if available
- Able to modify existing segmentation masks as well as creating new ones from scratch
- Issue #1 addressed: In absence of a preprocessing method, it still generates decent results. It is still highly encouraged to preprocess the data
- Issue #4 addressed: space in the folder name can exist
- Addressing fixed to be compatible with MacOS
v0.01
Create installation and usage instructions