ICAI DAY: Healthcare and AI

Join us on November 1, 2023, 12:00 – 17:00 for the ICAI Day Autumn Edition, exploring AI’s impact on healthcare. This event at Radboud University, Nijmegen, offers a deep dive into AI’s technical aspects in medicine.

The agenda includes insightful presentations (hybrid), engaging round table discussions (on site only), lead by experts in the field, and poster presentations of the ICAI Labs from across the Netherlands (on-site only). Highlights feature Nijmegen ICAI Lab’s cutting-edge work, AI in Radiation Oncology, MRI reconstruction challenges, and AI frameworks. Attendees can also explore ethical, governance, implementation, and legal aspects of AI in healthcare.

Registration

If you wish to attend the event please register using one of the following links:

• to attend the ICAI Day including the round table discussions please use the links found under each table description below (on-site).
• to attend the ICAI Day without participating in the round table discussions (both for online and in person attendance), register here.

Program

12:00 – 12:30 Registration

12:30 – 13:30 Round Table Lunch Discussion

13:30 – 13:40 Presentation on Overview of all ICAI health labs by Colin Jacobs, Assistant Professor and Principal Investigator within the Department of Medical Imaging of the Radboud University Medical Center.

13:40 – 14:05 Healthy AI Lab presentation by Henkjan Huisman, Professor of Medical Imaging AI at the NTNU  

14:05 – 14:30 CARA Lab presentation Towards AI-guided cardiac interventions by Jos Thannhauser, Assistant Professor within the departments of Medical Imaging and Cardiology of the Radboudumc.

14:30 – 15:00 Break & poster/ demo presentation

15:00-15:20 Brightlands Smart Health Lab presentation on ‘Navigating the future of AI in Radiation Oncology – Lessons learned and future prospects’ by Petros Kalendralis, Medical Physicist-Clinical Data Scientist. MAASTRO clinic

15:20-15:40 AI4 MRI Lab presentation on ‘Challenges and opportunities of AI-based MRI reconstruction’ by Thijs van Osch, professor in Radiology, experimental cerebrovascular imaging and he is vice-director of the C.J. Gorter Center for high field MRI

15:40-16:00 AI Challenges and frameworks – James Meakin, leads the Research Software Engineering team of the Diagnostic Image Analysis Group

16:00 – 16:05 Closing

16:10 – 17:00 Borrel/ Networking & posters

Round tables (on-site only)

The event will start with round table discussions. To attend the event in person, please register at one of the following tables:

• Table 1: Ethics in Healthcare A discussion on the moral and philosophical considerations surrounding medical practices, patient care, and decision-making in healthcare
• Register here.
• Table 2: Data Governance Exploring the strategies and policies for collecting, managing, and protecting healthcare data
• Register here.
• Table 3: Implementation of AI in Healthcare  Discussion on challenges and opportunities in integrating AI into various aspects of healthcare
• Register here.
• Table 4: Foundation models, e.g. LLMs for medical text Discussion on the use of large language models (LLMs) like ChatGPT in understanding and generating medical text, and their implications for healthcare
• Register here.
• Table 5: Explainable AI Discussing methods and techniques to make AI systems in healthcare transparent and interpretable, ensuring that decisions made by these systems can be understood by both professionals and patients
• Register here.
• Table 6: Legal aspects of AI in healthcare Analyzing the legal and regulatory landscape governing the deployment of AI technologies in healthcare, including issues related to liability, privacy, and compliance
• Register here.
• Table 7: Safety of AI in Healthcare Addressing the importance of safety measures, validation, and monitoring to minimize risks associated with AI applications in healthcare, ensuring patient well-being
• Register here.

Presentation Abstracts

Towards AI-guided cardiac interventions: An introduction to the CARA lab – Jos Thannhauser

In coronary artery disease, such as a myocardial infarction, percutaneous coronary intervention (PCI) is the cornerstone of treatment. During PCI, a series of crucial decisions are made by interventional cardiologists, including the assessment of the need for mechanical revascularization, the selection of procedural strategies, the choice of iterative treatment following stent placement, and stent optimization.

Optical Coherence Tomography (OCT) is a relatively new image modality with superior resolution compared to all other imaging techniques. Intracoronary OCT has the unique capability to accurately characterize the vessel wall and atherosclerotic disease processes, potentially aiding in procedural decision-making and stent optimization. However, its high-resolution output and the large amount of acquired images pose a time-consuming challenge for clinicians in real-time analysis.

The CARA lab is a collaborative ICAI lab of Radboudumc, Amsterdam UMC and Abbott, a global leader in medtech and OCT device development. As part of the LTP ROBUST program, the CARA lab develops AI solutions for automated analysis of OCT images in the catheterization lab, ultimately aiming for more efficiency and precision during cardiac procedures.

AI4MRI – Thijs van Osch – ‘Challenges and opportunities of AI-based MRI reconstruction’

MRI can be characterised as a multi-modal and relatively slow medical imaging modality. This makes MRI susceptible to motion artefacts and turns it into an expensive, expert-level imaging modality. The latter is especially limiting its application, since the safety-profile of MRI would make it suitable for much broader applications than the current usage late in the diagnostic work-up. Within this presentation, specific AI-based approaches will be presented to identify the level of motion present in an MRI-acquisition, leveraging of this information to improve reconstruction, as well as exploitation of the multi-modal aspects to accelerate MRI examinations.

“Navigating the future of AI in Radiation Oncology – Lessons learned and future prospects” by Petros Kalendralis

The field of Radiation Oncology stands at the cusp of a transformative era, driven by the integration of Artificial Intelligence (AI) into clinical practice. This presentation delves into the insights derived from work of the Clinical Data Science group of MAASTRO clinic and Maastricht University, shedding light on the progress made in the application of AI in Radiation Oncology.
We explore a variety of case studies that exemplify the impact of AI on improving patient care. These include but are not limited to:
FAIR data infrastructures: The transformation of data into a machine readable format for the implementation of federated learning studies.
Image Analysis: Explore the capabilities of AI in automating the analysis of radiological images, leading to faster and more precise diagnoses, such as the extraction of imaging biomarkers that can be useful for prediction outcomes.
Radiotherapy outcomes prediction: Learn about the predictive power of AI models in foreseeing patient response to radiation therapy, offering a personalized approach to treatment.
Radiotherapy planning Quality Assurance: Dive into studies showcasing how AI can streamline quality assurance processes, ensuring the safety and efficacy of radiation therapy treatments, such as the early detection of treatment planning errors.
This presentation highlights the promising aspects of AI in Radiation Oncology. It also explores the potential for AI to further revolutionize treatment planning, dosage optimization, and patient monitoring, ultimately advancing the field’s capabilities to deliver tailored and effective patient care.

Grand Challenge: Advancing Biomedical Imaging with Cloud-based Machine Learning Solutions by James Meakin

In this talk, we will explore the Grand Challenge platform, an open-source, cloud-based solution for advancing machine learning applications in biomedical imaging. Leveraging the power of Amazon Web Services (AWS), Grand Challenge offers a scalable, collaborative ecosystem for the end-to-end development and deployment of cutting-edge ML solutions. We will delve into the key features, capabilities, and advantages of Grand Challenge, and discover how we use it to bridge the gap between research and clinical practice.

“A scalable AI research platform for cloud-federated multi-center data ingestion, data curation, algorithm training and validation” We’ll talk about setting up large multi-center studies, having international readers, and teams of AI researchers experimenting with SOTA algorithms and bias free performance estimation of scientific and commercial AI to achieve trustworthy AI.