News and events archive

From the faculty

  • Title image

    2024-07-29 – 2024-09-02 2024-09-02

    XAI Summer School 2024

    From September 2 to 6, 2024, the Faculty of Informatics and the Institute of Computer Science at Masaryk University will host a summer school on Explainable Artificial Intelligence (XAI) in digital pathology.

    Key topics include the basics of pathology, AI research, data management and the application of AI in digital pathology. Practical training with RatioPath pipelines and RatioViz tools will culminate in a friendly competition showcasing diagnostic models and their explainability. The program will also emphasize best practices in handling sensitive medical data.

    GUESTS
    Attendees will gain foundational knowledge through lectures and practical sessions led by experts from Google (David Steiner), Charité (Norman Zerbe) and various universities or institutes (Petr Holub, Vít Musil, Rudolf Nenutil, Tomáš Brázdil at al.).

    PROGRAM
    Introduction into XAI

    • Pathology 101 for computer science students
    • What is explainability, why is it relevant for medical applications of AI

    Introduction into digital pathology - primary application use case of the summer school

    • Pathology 101 for computer science students
    • State-of-the-art of research in AI in digital pathology

    Data and their management in digital pathology

    • Overview of data generation for digital pathology
    • Annotations, management of data across research and health care (example of DICOM)
    • Imaging data
    • Practical examples how data is accessed for AI models
    • How data and outcomes of models is visualised
    • Clinical data and mining unstructured data
    • Handling sensitive data - principles requirements best practices
    • Practical trainings for students using projects on RatioPath pipelines with RatioViz

    Visualisation

    • Training of diagnostic model
    • Analysing explainability of the model
    • Potentially a small competition of teams on the same problem

    Web address
    Attachments