Version 0.5.0

What’s new in 0.5.0

  • App SDK is now compatible with MONAI v0.9.1 and later

  • DICOM Encapsulated PDF Writer. This built-in operator provides a simple and easy way to encapsulate PDF contents in a DICOM instance with or without referenced original DICOM instance(s)

  • DICOM Segmentation Writer with highdicom back end. This improves the description of segments with proper DICOM coding sequence(s)

  • Generated DICOM instances as AI evidence now have the attribute (0008,0201) Timezone Offset From UTC, in addition to the DICOM date and time which are set with values from the underlying operating system. The OS is expected to be synchronized with a known good timing source and have the correct timezone setting

  • Generated DICOM instance file names are now based on the SOP instance UID

  • Support DICOM instance level attribute matching in the DICOM Series Selection Operator

  • Operators and example applications are verified to be re-runable without needing to reinitialize the application object or re-load the AI model network. This will allow a main function or an external script to instantiate the application object once and use it to process multiple discreet inputs, either in a batch processing mode or in a long running service

  • Tutorials, in Jupyter notebooks, are re-organized and updated

  • Reference added for MONAI Deploy Express for hosting MAPs in development environments

  • Removed is the reference to the deprecated MONAI Inference Service

What’s fixed/updated

Please see the closed issues on Github and the closed pull requests on Github.

Known Issues

The example MedNIST Classifier application works but its MAP container image may fail to run.

Additional information

Please visit GETTING STARTED guide and follow the tutorials.

You can learn more about SDK usage through DEVELOPING WITH SDK.

Please let us know how you like it and what could be improved by submitting an issue or asking questions