Cognitive Science to Radiology: Using EEG and Eye-tracking to Determine Why, How, and When Novices and Radiologists Miss Signs of Breast Cancer in Multiple-abnormality Mammography and Tomosynthesis

Satisfaction of Search (SOS), where an abnormality is missed after a previous abnormality was found, is a
pervasive problem across different types of images, yet their impact on breast cancer detection is un-
known. Radiology and cognitive science have independently investigated why, how, and when SOS oc-
curs. The objectives of this grant are to bridge cognitive science research techniques with radiology to de-
termine: 1) the neural underpinnings of SOS, 2) the extent to which SOS affects breast cancer detection,
3) the extent to which advancements in imaging technology alleviates SOS, and 4) how expertise affects
SOS error rates. My central hypothesis is SOS contributes to breast cancer misses because a found ab-
normality becomes an attentional template. An attentional template is known in cognitive science as an
enhanced representation of a target maintained within memory and biases attention in search. The ra-
tionale for this project is when a first target is detected, it becomes an attentional template that utilizes and
biases attention necessary for detecting subsequent targets making searchers prone to SOS. However,
SOS may be mitigated: 1) by changing search strategies to counteract a first target becoming an atten-
tional template, 2) with advances in imaging technology, and 3) with expertise in breast cancer detection.
These influences need to be investigated to determine the extent to which they can alleviate SOS. This
work will pursue four specific aims to test the central hypothesis: 1) identify a mechanistic account for
SOS, 2) determine the extent to which SOS accounts for breast cancer misses, 3) compare SOS between
mammography and tomosynthesis, and 4) compare SOS error rates between different levels of expertise.
The research proposed is innovative because it will: 1) identify a mechanism responsible for SOS; 2) use
EEG and eye-tracking to study the neural basis of SOS; 3) use realistic, virtual breast images to investi-
gate the extent to which SOS affects breast cancer detection; 4) determine the extent to which the SOS
results from simplified search displays replicate in realistic-looking medical images; 5) be the first time
SOS is investigated with 3D imaging; and 6) determine the extent to which SOS rates vary by expertise in
breast cancer detection. The proposed research is significant because: 1) it will develop a neural under-
standing of how SOS contributes to cancer detection; 2) determine how SOS varies across different
search modalities and by expertise; and 3) ultimately improve breast cancer detection by determining how
these influences impact SOS rates. To complete the proposed research, in pursuit of his long-term goal of
becoming an independent researcher at a Tier-I institution, the PI will train using EEG and eye-tracking.
His training will include auditing graduate courses, attending workshops, weekly meetings/training with his
mentors, and bi-annual assessments of his EEG/eye-tracking expertise with his mentors.

Grant Number: 5R00CA267163-04
NIH Institute/Center: NIH
Principal Investigator: Stephen Adamo