2nd Hybrid Training School
„Radiomics and AI in Molecular Imaging””

The 2nd Training School on “Radiomics and AI for Molecular Medicine” will be held from October 11 to 13, 2021. The school format will be hybrid and sessions will be streamed from the studio of the Medical University of Vienna. The school will entail presentations and hands-on and leave ample space for Q&A. Building on the 2020 School , it will first rehearse the basics of Radiomics and AI before giving the floor to expanded perspectives on data curation, pre-processing, concepts of machine learning and model validation. The theoretical perspectives will be embraced by statements on pre-/clinical use-cases. Hands-on sessions will address various open-source and zero-code, minimum-tune solutions for radiomics and machine learning. For details see here: https://www.comulis.eu/training-school-radiomics-and-ai . We kindly acknowledge the generous support by the COST Action “COMULIS”.

Target audience
This course aims at imaging stakeholders who are interested in extracting levels of information beyond plain 3D or 4D image analysis. This includes imaging and medical specialists, imaging physicists, as well as early-stage computer scientists with a strong interest in radiomics, ML and data processing. We anticipate little prior experience, and no programming skills are needed to attend the course. Fellows will be trained in good scientific practice in Radiomics/AI/Imaging, in the core and mid-level aspects of machine learning, data preparation and model building and validation using coding-free tools.

Key Objectives

• To highlight the basics of Radiomics and AI in the context of image-guided diagnosis and therapy management using both pre-/clinical data.
• To appreciate the need for high quality input data to AI, and to learn about data curation prior to feeding AI algorithms.
• To deepen knowledge of machine learning and AI to medical imaging.
• To learn about strategies to validate prediction models.
• To learn about Open Access tools, zero-tune and limited/zero-code algorithms to extract radiomic features and build prediction models.

• The role of Radiomics and AI in clinical applications
• Planning clinical research activities in the world of AI
• Feature extraction
• Data preparation and accounting for data imbalances
• Data analysis, harmonization, and the curse of dimensionality
• Model interpretation and validation

Our faculty is composed of international experts in the fields of molecular imaging, computer sciences and AI, and includes – thankfully again -