Abstract
Due to resistance amongst professionals to adopt Artificial Intelligence (AI) technologies in radiology, there is a need to shift focus to end-user involvement in design processes, to enable trust. This body of work investigated preferences of chest radiograph AI User Interfaces (UI) amongst radiographer and radiologist end-users, including radiographers and radiologists. A scoping review explored existing literature on the preferred AI UIs in radiology for end-users, though found limited research in this area. A survey explored preference of three existing chest radiograph AI UIs (a) Heat Map (b) Area of Interest (c) Textual Report (d) Combination of a and c (e) Combination of b and c, against System Usability Scale (SUS) questions, and identified preference for b, c and e. An eye-tracking study with a Think-Aloud element explored eye gaze metrics for participants reviewing (i) Heat Map, (ii) Area of Interest (iii) Textual Report (iv) Abnormality Score (v) Combination UIs. A Friedman test identified significance between the number of fixations and chest AI UIs design (χ² (4) = 16.9, p=0.002). Among the AI UIs in this study, the Heat Map was the least preferred due to perceived ambiguity. The Area of Interest UI was more favourably received for its clarity, and the Textual Report UI was appreciated though some found the terminology used distracting. To understand barriers and enablers to AI integration more broadly, a multidisciplinary AI study conference was hosted and focus groups divulged themes such as fear of overreliance and a desire for phased technology implementation. A Designathon workshop contributed to creation of a chest radiograph AI UI prototype with features including the use of dotted lines to outline pathologies, AI toggling features and traffic light colouring. The final usability survey of the prototype, identified a perceived SUS score of 75, exceeding the literature average.Due to copyright restrictions, Chapter 2 is redacted from view for two years. The full unredacted version is embargoed until 30 September 2028.
| Date of Award | Sept 2026 |
|---|---|
| Original language | English |
| Sponsors | Department for the Economy (Northern Ireland) |
| Supervisor | Ciara Hughes (Supervisor) & Sonyia McFadden (Supervisor) |
Keywords
- artificial intelligence
- radiology
- radiography
- imaging
- chest radiography
- user interface
- human-computer interaction
- trust
- integration
Cite this
- Standard