Guest talk with Dr. Nora Castner

October 17, 2024

In medicine, human-AI collaboration can promote diagnostic performance. Expert gaze models can offer insight into how professionals adapt their behavior to AI. Moreover, improving interaction design of medical AI-systems improves understanding and trust. This talk will provide an overview of expert gaze behavior while interacting with a medical AI support system and investigate diagnostic visual strategies and system usability. We combine two fields in eye tracking, expert gaze behavior during medical image examination and gaze for usability evaluation. Thus, we started with: How do experts scanpaths relate to bitewing inspection for caries detection? Then, how does an expert’s scanpath deviate from their usual approach due to AI? Ultimately, bringing us to the question: Does AI-support potentially disrupt typical, expert scanpaths? With this outlook, we can avoid potential negative implications on workflows and fatigue by helping design these systems. Ultimately, this knowledge of expert gaze features can aid developing proficient diagnostic workflows and seamless interaction with medical AI systems.

Nora Castner is a post-doctoral researcher in digital visual healthcare in human centric AI at the ZEISS Vision Science Lab, which is an Industry on Campus Collaboration. Her current work investigates ways to promote a lean communication system with user and medical AI systems. Her focus is on visual search modeling and how it promotes the interdisciplinary context of machine learning, with the underlying question of how AI can transform disciplines. Nora Castner has been working with eye tracking technology since 2016 and stays involved in this community as a committee member of the eye tracking related conference (ETRA).