Case Study on Improving Image Annotation Efficiency in Digital Pathology

Robert Prikryl

Supervisor(s): Ing. Miroslav Laco, PhD.

Slovak Technical University

Abstract: Medical image annotation is a crucial component in the creation of high-quality datasets for AI applications in histopathology. This study investigates whether usability and efficiency can be improved through interface simplification, task-oriented configuration, and AI-supported workflows. To explore this, existing digital histopathology annotation tool AnnotAid was modified using a modified Double Diamond design methodology. The first prototype introduced configurable interfaces, multi-format image support, and a Wizard-of-Oz-based AI simulation to emulate AI-generated outputs. In the second iteration, the issues identified during the initial evaluation, such as missing onboarding and AI explainability, were addressed in a newly improved prototype version. The first prototype was evaluated through usability testing with five non-medical participants in domain-shifted settings. Both quantitative metrics and qualitative observations were used to identify key usability issues. A second evaluation is planned with a histopathology domain expert using the modified prototype developed in the second iteration.
Keywords: Human-Computer Interaction
Full text:
Year: 2026