Comparison Display Resolution On User Impact For Digital Pathology

Abstract

Introduction/ Background

Digital pathology images are very large, up to 100000x100000 pixels which are 30 to 50 times larger than a radiological image for which 12 Mega Pixels (MP) medical displays can be used. Higher resolution displays may have an important influence on digital pathology ergonomics. Three displays with varying resolutions were studied to determine their impact on user interaction.

Aims

Our hypothesis was that “with higher resolution displays, pathologists need less interaction such as panning and zooming actions and can focus more on image contentâ€. A psycho-physical study has been carried out for validating this hypothesis at the University of Pittsburgh Medical Center.

Methods

Three experienced pathologists were selected. Seventy pathology including a wide variety of histological and cytological diagnoses were digitized (Aperio Scanscope XT scanner) and used in a previous study [Ava15]. Customized and optimized viewing software was used to display images and record pathologist’s interactions such mouse clicks, zooming and panning. Three medical displays with different different resolutions were used: 2MP (BARCO MDSC-2124), 4MP (BARCO MDPC-4130) and 12MP (BARCO MDCC-12133), all with the same maximum luminance. Scripts were used for statistical analyze and 1D, 2D, 3D plotting results. User interactions with each image were used to recreate videos documenting of their exact navigation with each digital slide.

Results

The results of number of zooming and panning interactions are given in the Table 1, as well as averages. When display resolution was increased, the number of panning and zooming interactions significantly decreases for all three observers. For panning, there was on average 1172 panning actions for the 2MP and 951 actions for the 12MP display. For zooming actions, there was on average 12315 zoom actions for the 2MP and 2847 actions for the 12MP display. Between the 2MP and 12MP displays, the ratio of the number of zooms was 4:1 in favor of the 12MP monitor. On figure 1, the 3D plots of one case for the three monitors show the navigation through the slide and show lesser points for higher resolution display. With higher resolution the pathologist goes more directly to the Region Of Interest (ROI) for making the decision. Figure 2 shows more analysis of the zoom values across the cases for the three monitors. The pathologists have the tendency to remain close to a value of 1 with the 12 MP display where a value of 1 means that no zoom is applied. This is illustrated by the Figure 2 showing for observer 2 the boxplots of the zoom values for the three displays. It clearly shows that with higher resolution display the trend goes to get closer to one for the zoom value meaning no need to zoom in in the image. We used three different displays instead of one unique display with three different resolutions. Though using just one display would have reduced variability of differing LCD panels, pixel size and structure, it would not have been commercially or clinically realistic.

Despite the limited number of pathologists, this study shows that display resolution used for digital pathology is important. Higher resolution monitors significantly help reducing the number of user interactions and thereby can minimize pathologist fatigue when reading digital slides.