Related Experiment Video
Updated: Jul 6, 2026
05:33
Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
Published on: July 11, 2025
Formative Usability Testing of Artificial Intelligence in Pathology: The Challenge of Assessing Acceptability
Natasha Alvarado1,2, Derek Magee3, Darren Treanor3,4,5
1Centre for Digital Innovation in Health and Social Care, University of Bradford, Bradford, UK.
Studies in Health Technology and Informatics
|May 17, 2025
Summary
An AI tool for digital pathology assists with skin cancer reporting. Early testing revealed usability issues and AI inaccuracies, highlighting the need for accuracy to ensure efficiency and clinician trust for successful adoption.
Area of Science:
- Digital Pathology
- Artificial Intelligence in Medicine
- Oncology Diagnostics
Background:
- Digital Pathology enables AI integration for enhanced diagnostic and reporting capabilities.
- AI tools are being developed to analyze digital Hematoxylin and Eosin (H&E) stained tissue images for skin cancer cases.
- The objective is to pre-populate pathology reports, aiming to improve pathologist efficiency and save time.
Purpose of the Study:
- To assess the ease of use and acceptability of an initial iteration of an AI pathology assistant.
- To identify usability issues and gather feedback for future AI tool development in digital pathology.
Main Methods:
- A think-aloud evaluation was conducted with twelve pathologists across seven UK hospitals.
- Participants completed a pathology report using the AI tool, followed by a brief interview.
- Qualitative data from think-aloud sessions and interviews were analyzed to identify usability challenges.
Main Results:
- The think-aloud evaluation identified several issues impacting the AI tool's ease of use.
- AI performance, specifically inaccuracies in populating report items, hindered the assessment of tool acceptability.
- AI inaccuracies introduced additional tasks, potentially decreasing overall reporting efficiency.
Conclusions:
- AI accuracy is crucial for evaluating the integration of AI tools into clinical workflows and improving efficiency.
- Clinician trust in AI performance is essential for the successful adoption of these tools in practice.
- Further development is needed to refine AI accuracy and usability for effective integration into digital pathology reporting.
