Abstract
Radiologists spend prolonged periods inspecting and interpreting medical images, and thus may adapt to the images as they are scanning them. In previous work we examined potential adaptation effects in the context of images acquired from full-field digital mammography (FFDM), and showed that these images can induce a variety of visual aftereffects including changes in contrast sensitivity and perceived texture (Kompaniez et al. 2013, 2018). Screenings are increasingly utilizing digital breast tomosynthesis (DBT), which unlike FFDM uses 3D imaging to produce multiple images at different depth planes, providing improved diagnosis. Images from the two modalities have textural differences and during a screening both may present and visually compared. We examined whether similar patterns of textural aftereffects occur across the different imaging modalities and also tested for transfer of adaptation between them. Stimuli were sections from FFDM and DBT (single plane) images of the same normal tissue that had BIRADS classifications of dense or fatty. Observers (non-radiologists) adapted to pairs of images from one category and then judged the perceived texture of image composites formed by averaging across categories. During adaptation images were rotated or mirrored at 200 ms intervals to reduce local light adaptation. Adaptation to dense images caused blends to appear more fatty and vice versa, and these effects occurred for both modalities. Adaptation to an image of the same tissue from one modality also biased the appearance of the tissue imaged with the other modality, though these effects were weaker when images from the two modalities were more similar. Similar interactions were also observed in a spatial analog of the adaptation, such that image blends tended to appear less like the images they were surrounded by. Our results suggest that both temporal and spatial contrasts could influence visual judgments within and across these different medical imaging platforms.