Skip to content

Latest commit

 

History

History
192 lines (178 loc) · 8.46 KB

CHANGELOG.md

File metadata and controls

192 lines (178 loc) · 8.46 KB
Raw

Changelog

All notable changes to MONAILabel are documented in this file.

The format is based on Keep a Changelog.

Unreleased

0.6.0 - 2022-12-19

Added

  • Pathology Models
  • QuPath Extension: 0.3.1
    • User experience enhancements
    • MONAI Label specific Toolbar actions
    • Drag and Drop ROI to run auto-segmentation models
    • Single click to run interaction models (NuClick)
    • Support Next Sample/ROI for Active Learning
  • Experiment Management
  • 3D Slicer: Detection model support in MONAI Bundle App for Radiology use-case
  • Multi-GPU/Multi-Threaded support for Batch Inference

Changed

  • Support the latest version of bundles in MONAI Bundle App
  • Upgrade vertebra pipeline
  • MONAI version >= 1.1.0

Removed

  • DeepEdit model for Nuclei segmentation in Pathology

0.5.2 - 2022-10-24

Added

Changed

  • Support for MONAI 1.0.1
  • Option to disable DICOM to NIFTI conversion for OHIF usecase
  • Proxy URL fixes for GCP + DICOMWeb support
  • 3D Slicer UI Improvements
  • Improvements on MONAI Bundle App
    • support local bundles (pre-downloaded)
    • support customized scripts

Fixed

  • Fixes for multi label output of segmentation_nuclei model (Pathology)

Removed

  • Remove option to run ALL Training Tasks in 3D Slicer

0.5.1 - 2022-09-16

Added

  • Endoscopy Sample App
    • Tool Tracking segmentation model | Demo
    • InBody vs OutBody classification model | Demo
    • DeepEdit interaction model
    • CVAT Integration to support automated workflow to run Active Learning Iterations
  • Multi Stage vertebra segmentation in Radiology App

Changed

  • Improving performance for Radiology App
    • Support cache for pre-transforms in case repeated inference for interaction models
    • Support cache for DICOM Web API responses
    • Optimize pre transforms to run GPU for max throughput
    • DICOM Proxy for wado/qido
  • Improvements for Epistemic (v2) active learning strategy
  • Support for MONAI 1.0.0 and above

Fixed

  • Scribbles to support MetaTensor

Deprecated

  • TTA Based active learning strategy is deprecated

0.4.2 - 2021-07-25

Added

Changed

  • Support for MONAI 0.9.1

0.4.1 - 2021-07-05

Changed

  • MONAI dependency version to 0.9.0

0.4.0 - 2022-06-13

Added

  • Pathology Sample App
    • DeepEdit, Segmentation, NuClick models
    • Digital Slide Archive plugin | Demo
    • QuPath plugin | Demo
  • Histogram-based GraphCut and Gaussian Mixture Model (GMM) based methods for scribbles

Changed

  • Support for MONAI (supports 0.9.0 and above)
  • Radiology Sample App (Aggregation of previous radiology models)
    • DeepEdit, Deepgrow, Segmentation, SegmentationSpleen models
  • NrrdWriter for multi-channel arrays
  • Static checks using pre-commit ci

Fixed

  • 3D Slicer
    • Support Segmentation Editor and other UI enhancements
    • Improvements for Scribble Interactions
    • Support for .seg.nrrd segmentation files
    • Support to pre-load existing label masks during image fetch/load

Removed

  • SimpleCRF and dependent functions for scribbles

0.3.2 - 2022-02-28

Added

  • SSL and multiple worker options while starting MONAI Label server
  • Scribbles support for OHIF Viewer
  • Add Citation page

Changed

  • Support for MONAI (supports 0.8.1 and above)
  • Upgrade PIP Dependencies

Fixed

  • Load pretrained models for TTA and Epistemic Scoring
  • Load MMAR API for deepgrow/segmentation apps
  • Documentation Fixes

0.3.1 - 2021-12-29

Added

  • Flexible version support for MONAI (supports 0.8.* instead of 0.8.0)

Changed

  • Strict Flag is set to False while loading pretrained models

Fixed

  • Inverse transform for DeepEdit sample app
  • Documentation Fixes

0.3.0 - 2021-11-28

Added

  • Multi GPU support for training
    • Support for both Windows and Ubuntu
    • Option to customize GPU selection
  • Multi Label support for DeepEdit
    • DynUNET and UNETR
  • Multi Label support for Deepgrow App
    • Annotate multiple organs (spleen, liver, pancreas, unknown etc..)
    • Train Deepgrow 2D/3D models to learn on existing + new labels submitted
  • 3D Slicer plugin
    • Multi Label Interaction
    • UI Enhancements
    • Train/Update specific model
  • Performance Improvements
    • Dataset (Cached, Persistence, SmartCache)
    • ThreadDataloader
    • Early Stopping
  • Strategy Improvements to support Multi User environment
  • Extensibility for Server APIs

Changed

  • Operate histogram likelihood transform in both normalized and unnormalized modes for Scribbles

Removed

  • DeepGrow Left-Atrium

0.2.0 - 2021-09-23

Added

  • Support for DICOMWeb connectivity to local/remote PACS
  • Annotations support via OHIF UI enabled in MONAI Label Server
  • Support for native and custom scoring methods to support next image selection strategies
    • Native support for scoring and image selection using Epistemic Uncertainty and Test-time Augmentations (Aleatoric Uncertainty)
  • Custom ScoringMethod and Strategy implementation documentation
  • Scribbles-based annotation support for all sample apps

Changed

  • Previously named generic apps now have default functionality under deepedit, deepgrow and segmentation
  • Updated Modules Overview documentation to include interaction between ScoringMethod and Strategy

Removed

  • All spleen segmentation sample apps (DeepGrow, DeepEdit, auto-segmentation)

0.1.0 - 2021-07-14

Added

  • Framework for developing and deploying MONAI Label Apps to train and infer AI models
  • Compositional & portable APIs for ease of integration in existing workflows
  • Customizable design for varying user expertise
  • 3DSlicer support
  • Support for multi-label auto-segmentation
  • Template apps to customize the behavior of DeepGrow and DeepEdit
  • Automated segmentation of left atrium, spleen
  • DeepGrow AI annotation of left atrium, spleen
  • DeepEdit AI annotation of left atrium, spleen