ABSTRACT

AI methods will revolutionize all areas of diagnostic medical imaging in the coming years. Complexity of case workup is also continuously increasing, in order to meet the requirements of targeted therapy and of immuno-oncology in particular. In addition to classical morphologic and microscopic interpretation, diagnoses are increasingly based on complex molecular information, such as from “omics” fields. Cancer screening programs are expanding and there is a growing shortage of specialists.

The aim of EMPAIA is the establishment of an ecosystem for image-based, AI-assisted, diagnostic services, using pathology as an example. By creating a standardized marketplace within a clearly defined legal framework, physicians will be able to routinely use validated and approved AI solutions.

In EMPAIA AT our group will make machine results transparent, traceable and thus interpretable for a medical expert. Pathologists will thus be able to question an AI decision in a specific context as to why. For this purpose, we will provide methods and approaches for the determination of explanatory quality and causality. Ultimately, our research results will support the development of novel Human-AI interfaces, which support an efficient and ethically responsible use of AI in pathology.

Link to the main project website:
https://empaia.org

  • Project period

    2020– 2023

  • Keywords

    Human-Centered AI, Human-AI Interfaces, Human-AI interaction, Explainability, Causability