Patrick Brennan

Ph.D.

University of Sydney

Professor

NLST (Learn more about this study)

NLST-333

Jul 31, 2017

Gist response

This work aims to investigate the role of the gist of the abnormal in detection of lung cancer in low-dose CT images. The gist of the scene refers to the perceptual information that an observer receives from a momentary glimpse of image. It is a remarkable characteristic of the human visual system which helps us to intelligently allocate attention in a new environment even if a complex scene is presented for only 20 milliseconds. In the context of medical image interpretation, the gist response refers to instantaneous perception of changes relating to an abnormality in an image, which can be inappropriately cancelled out following more detailed image evaluation. Previous studies have been shown that the radiologists are able to detect the abnormality in chest x-rays and mammograms based on less than half-second presentation of the image. It has been previously hypothesized that visual search starts with a global (gist) response that establishes context, identifies gross deviations from known normal pattern [1], and initiates a sequence of foveal checking fixations, however, Evans et al [2, 3] also suggests that categorizing abnormal cases based on extracting a gist whilst not necessarily being constrained by decisions from subsequent foveal fixations may be of some diagnostic utility. The work by Evans et al work [2] showed that radiologists had an above chance performance for detecting subtle abnormalities at a range of stimulus durations (250–2000 milliseconds) even when their ability to localize these abnormalities (with no time restrictions) was at chance levels. A subsequent study also showed that mammograms with no visible signs of cancer, which have been reported as normal by radiologists, had an above chance of being considered abnormal following the gist when these images came from women with previously reported cancer in the contra-lateral breast [3]. This evidence suggests that the previous mammogram with no visible overt signs of cancer may still contain important predictive information that should not be ignored, and may be identified by exploiting the gist phenomenon. In this work we aim to investigate whether the radiologists have above-chance level performance in detecting cancer cases from normal cases in screening low-dose CT images. Also, this work aims to establish whether expert radiologists have an above-chance performance in distinguishing CT images of the participants who diagnosed with cancer at subsequent screening from CT images arising from participants who have never been reported with lung cancer in the study period.
[1] Kundel HL, Nodine CF (1975) Interpreting chest radiographs without visual search. Radiology 116(3):527–532.
[2] Evans KK, et al. (2013) The gist of the abnormal: Above-chance medical decision making in the blink of an eye. Psychonomic bulletin & review 20 (6):1170-1175.
[3] Evans KK, et al. (2016) A half-second glimpse often lets radiologists identify breast cancer cases even when viewing the mammogram of the opposite breast. Proceedings of the National Academy of Sciences 113(37):10292-10297.