The research of the MERAI Lab is about developing AI-supported software solutions for healthcare to improve the accuracy of imaging interpretation in the lung oncology field, reduce the time needed to report scans, and improve the cost-effectiveness of the healthcare system. MERAI Lab is a collaboration between Radboud UMC and MeVis Medical Solutions.

‍

Healthcare costs are globally rising. The workload of radiology departments has substantially increased and is still increasing, and, as a result, radiologists are under large pressure and risk of burn-out. The imminent implementation of lung cancer screening and the rapid increase in the availability of novel cancer treatments such as immunotherapy will result in a large increase in imaging. One key technology which has the potential to reduce the workload of radiologists is artificial intelligence. Simple, repetitive tasks can be taken over by AI algorithms, leaving more difficult and context-dependent differential diagnosis questions to radiologists and putting more focus on reporting to patients (human-centric care). In addition, we foresee a demand and chance for trained human readers, not necessarily medical specialists, aided by AI to prescreen images before being sent to radiologists. Ultimately, autonomous AI will be able to triage screening scans such that only abnormal studies and a subset of studies about which AI is uncertain need to be interpreted by human experts.

‍

The MERAI Lab is focused on creating world-leading AI-supported software solutions for healthcare. Our mission is to develop AI algorithms that can improve the accuracy of imaging interpretation in the lung oncology field, reduce the time needed to report scans, and improve the cost-effectiveness of the healthcare system. To ensure the responsible use of these algorithms, the MERAI Lab aims to create robust and trustworthy AI solutions that perform at a level close to human experts. The lab will also evaluate and validate the AI technology it develops, using a global network of hospitals to assess its impact on healthcare.

‍

To improve the accuracy of AI and facilitate its adoption in society, the MERAI Lab will employ a number of strategies. These include using human-in-the-loop annotation strategies to build large, well-curated databases for training, working with clinical stakeholders to optimize the use case of each developed algorithm, and considering ethical, legal, and societal factors to stimulate successful adoption. The lab will also focus on resilience, specifically through the development of efficient out-of-distribution detection methods to cope with distributional shifts. By implementing these strategies, the MERAI Lab hopes to create AI solutions that are reliable and effective in improving healthcare outcomes.