Description
Title:An analysis of User feedback of different interfaces and designs for clinically relevant decision support when using AI: what do imaging professionals’ prefer?
Short Title: User Friendly AI study
Purpose:
The aim of the study was to investigate the sensemaking and cognitive behaviour of healthcare staff when interacting with Explanation User Interfaces (EUI) and gather their user preferences of chest radiograph Artificial Intelligence (AI)-based EUIs.
Methods/background:
A large part of human and machine interaction involves the EUI that clinicians use as the connection between medical diagnosis or report. However, there is currently a lack of EUI standardisation within medical imaging and AI (Schalekamp, Klein and van Leeuwen, 2022).
To build on the findings from an international questionnaire undertaken at ECR 2024, a mixed methods study was undertaken incorporating eye-tracking, Think-Aloud and questionnaire. Diagnostic radiographers’, radiologists’, trainee radiologists’ and student radiographers’ visual preferences when reviewing 4 different types of chest radiograph AI EUIs, were investigated within this research (salience maps, textual reports, area of interest and abnormality score EUIs). Participants were asked to review the images whilst wearing eye-tracking software and say what they were thinking i.e. the “Think-Aloud” method. The post study questionnaire asked the participants about their perceived level of confidence against the four different interfaces.
Results/findings:
24 participants took part in the study which allowed an understanding of which components of the chest radiograph EUI are focused on and subsequently preferred. Descriptive statistics on the eye-tracking data relating to fixations and saccades indicating where the viewer gave most attention identified interesting patterns in participants. Insights on participant preference were further detailed from the Think-Aloud and post-study questionnaire data.
Conclusion:
Understanding user preference for chest radiograph AI EUI is important to ensure appropriate user engagement with the information provided by the technology and in turn give radiographers and radiologists the ability to explain this to patients.
Limitation:
• Small sample size may have affected the generalisability of findings of this research.
• Limitations of the eye-tracking software
| Period | 2025 |
|---|---|
| Event title | European congress of Radiology |
| Event type | Conference |