Learn About AI Model Interoperability with HALO AI
Summary
As artificial intelligence becomes increasingly integrated into digital pathology, interoperability between image analysis platforms, externally developed models, and custom computational workflows is becoming increasingly important. This webinar will demonstrate multiple approaches for integrating AI models with HALO®, highlighting how users can move models into, out of, and alongside the HALO ecosystem.
First, we will demonstrate how externally developed models can be imported into HALO using the ONNX standard, allowing models created in external machine learning environments to be incorporated into HALO-based analysis workflows. Next, we will show how a model developed using HALO AI can be exported to ONNX and used for inference in an external Python workflow, illustrating how HALO-trained models can be extended beyond the platform.
Finally, we will demonstrate a service-based integration in which a custom AI model is deployed in a Docker container and connected to HALO through the HALO API. This approach enables models to remain within their native software environment while still participating in HALO-driven analysis workflows.
Together, these examples illustrate different levels of model portability and workflow interoperability, providing flexible options for integrating open-source, internally developed, and third-party AI models into digital pathology workflows.
Learning Objectives
- Explore approaches for enabling AI model interoperability with HALO.
- Import ONNX models into HALO and incorporate them into digital pathology analysis workflows.
- Export HALO AI-developed models to ONNX for use in external Python-based workflows.
- Describe how the HALO API can connect containerized or externally deployed AI models with the HALO ecosystem.
Presenter
Donald Allen
Product Manager, HALO AI
Indica Labs
Donald Allen earned his BS in Microbiology, Immunology, and Molecular Genetics from UCLA. He has over 5 years of experience in R&D with Dako/Agilent in the field of companion diagnostics. He is experienced in feasibility of Class III medical device development, deployment of clinical trial assays to CROs, and the development of manual scoring algorithms. Donald now works in the field of digital pathology and image analysis to combine his work experience and his passion for computers and technology. He joined Indica Labs as an Applications Scientist in 2017, was promoted to a Sr. Applications Scientist in 2020, and most recently was promoted to the role of HALO AI Product Manager.
