Abstract
Humans are very adept at extracting the "gist" of a scene in as little as a fraction of a second. This "gist" perception can be learned for novel stimuli. We have found that this extends to the gist of breast cancer. Radiologists can distinguish between normal mammograms and those containing subtle signs of cancer at above chance levels with 250 msec exposure but are at chance in localizing the abnoramility. This pattern of results suggests that they are detecting a global signal of abnormality. What are the stimulus properties that might support this ability? Across four experiments, we systematically investigated the nature of the "gist" signal by asking radiologists to make detection and localization responses about briefly presented mammograms in which the spatial frequency, symmetry and/or size of the images was manipulated. Interestingly, the signal is stronger in the higher spatial frequencies. Performance is poor with low-pass filtered images but almost as good with high-pass as with unfiltered images. Performance does not depend on detection of breaks in the normal symmetry of left and right breasts. Moreover, above chance classification is possible using images of the normal breast of a patient with overt signs of cancer in the other breast. Some signal is present in the portions of the parenchyma (breast tissue) that do not contain a lesion or that are in the contralateral breast. This signal does not appear to be a simple assessment of breast density. The experiments indicate that detection may be based on a widely-distributed image statistic, learned by experts (Non-expert observers perform at chance). The finding that global signal related to disease can be detected in parenchyma independent of the appearance of the cancer may have further relevance in the clinic.
Meeting abstract presented at VSS 2016