A Novel Eye- and Head-Tracking Interface for Digital Pathology Reduces Hand Movement and Improves Target Detection...

The transition toward digital pathology brings ergonomic and efficiency challenges with mouse or other hand-based input devices for digital slide assessment, requiring users to move their hands frequently, raising the risk of repetitive strain injury. This work explores non-hand-based input devices for digital pathology.

To determine the feasibility of using eye and head tracking as input devices for completing representative tasks in digital pathology, compared to the computer mouse.

Eight pathologists completed 5 representative digital pathology tasks: measuring, counting, zooming, panning, and target identification. Each task was first completed with the computer mouse as the input device, then repeated with eye and head tracking as the input devices, after a washout. Eye gaze data were collected throughout. Pathologists were asked to complete a survey on their experience.

Measurements made with eye gaze were 0.18% more accurate than those made with mouse (P = .001). Differences in point localization accuracy for measurement and counting with eye gaze were both less than 10 μm, compared to mouse (P < .001, for both). Hand movement was reduced for counting, zooming, and panning with eye and head tracking (P = .04, P = .009, and P = .04, respectively). No targets were missed with head tracking, whereas 8 targets were missed with mouse. Most pathologists favored eye tracking for counting and zooming.

Eye and head tracking are feasible input devices for digital pathology and provide comparable accuracy to the computer mouse. Head tracking suggests continuous panning of the image is a superior and safer method when an exhaustive search for targets is required.