Skip to content

Releases: Project-MONAI/monai-deploy-app-sdk

v3.0.0

23 Apr 01:28
Compare
Choose a tag to compare

Version 3.0.0 (April 22nd, 2025)

What's new in 3.0.0

  • This version of the App SDK is based on the newly released Holoscan SDK v3, and is expected to be so with future minor releases of Holoscan SDK v3.

  • Starting with version 3.0.0, Holoscan SDK and Holoscan CLI are released in separate packages, and as such, this version of the MONAI Deploy App SDK has both as dependencies. As of now, version 3 of both packages are compatible.

  • Remote inference on Triton Inference Server is now supported. Effort was made to make it user-friendly so existing example applications can be easily converted to use this feature by simply providing the network location of the server as well as the Triton model configuration file sans the actual model files. An example application has been provided to demonstrate such use case.

Key changes

  • Cincinnati Children's Hospital Medical Cente researchers, @Bryan @Will and Elan, contributed numerous enhancements from experience developing and deploying MONAI based AI applications in clinical environments, to name but a few

    • Enhanced the DICOM data loader to handle multi-phase DICOM series where multiple acquisitions exist and some DICOM SOP instances have the same image pospositiontion patient.

    • Enahnced the DICOM series to volume operator to better handle the data types of the converted volume image for improved efficiency and memory usage.

    • Enhanced the DICOM Segmentation operator to populate DICOM tags with AI model information which are consistent with other DICOM writers in the SDK.

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

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