Designing a digital pathology workstation for routine practice

Sammanfattning: The role of the pathology lab is important in the future of cancer care. In order to further personalize the care for cancer patients, more precise review of tumor specimens is needed in order to guide clinicians between different treatment strategies. New digital imaging technologies is one promising possibility that might allow pathologists performing more and better work with the same amount of resources. Early scanning systems and workstations have been shown to be inefficient and have not met the pathologists’ needs, who still perform most of their diagnostic work with mechanical microscopes. In this thesis, we analyze the pathologist's work with early digital workstations and present a set of new solutions in order to increase the performance of the interaction with these systems. First, we review the implementation process of two current digital systems in two pathology labs in Sweden (Paper I), followed by study of the navigation behavior that is performed by pathologists when they explore large digital slides of cancer specimens (Paper II). With a specific focus on design solutions that work within medical routine practice, three different input devices for navigation in large images was compared with pathologists as participants (Paper III), as well as a visualization technique, inspired by semantic zoom in order to facilitate certain tasks for pathologists (Paper IV). The results provided in this thesis points towards the same conclusion that have made in other domains: When good usability engineering is combined with technological advances, this can make novel technology become useful for real. For a Human-Computer Interaction (HCI) researcher, the pathologist case represents an especially demanding use of zoomable user interfaces. This has driven us to enhance efficiency of such interfaces further in order for them to become useful. The research findings offered within this thesis are particularly important to the field of digital pathology. However, our findings could also have a bearing on the design of zoomable user interfaces.