TY - JOUR
T1 - Do different radiologists perceive medical images the same way? Some insights from Representational Similarity Analysis
AU - Hegdé, Jay
AU - Bart, Evgeniy
N1 - Funding Information:
We thank Ms. Fallon Branch for help with the preparation of this manuscript, and Ms. Jennevieve Sevilla for excellent technical assistance, and the National Cancer Institute (NCI) for funding our participation in the HVEI panel on medical image perception. We are grateful to the NCI and to Dr. Jeremy Wolfe of Harvard Medical School for organizing the aforementioned RSNA testing facility each year. This study was supported by Army Research Office (ARO) grants W911NF-11-1-0105 and W911NF-15-1-0311 to JH.
Publisher Copyright:
© 2019, Society for Imaging Science and Technology
PY - 2019/1/13
Y1 - 2019/1/13
N2 - Characterizing what experts perceive in medical images is a difficult problem, both because doing so requires somehow characterizing the internal mental representations of the observer, and because the underlying diagnostic information tends to be abstract and not readily describable in terms of well-defined image features. Representational Similarity Analysis (RSA) is a method originally developed in mathematical psychology that provides a theoretically sound and quantitative framework for measuring the mental representations of visual images in human observers. Here we used RSA to measure the extent to which the same underlying set of mammograms elicit similar mental representations in different practicing radiologists (N = 26). We found that the internal representations were statistically indistinguishable across different radiologists (p > 0.05). Moreover, the mental representations significantly parallel the diagnostic information in the images (p < 0.05 for each subject), indicating that various radiologists perceived the same set of diagnostic information in the underlying images. Together, these results indicate that medical images elicit similar mental representations in different radiologists.
AB - Characterizing what experts perceive in medical images is a difficult problem, both because doing so requires somehow characterizing the internal mental representations of the observer, and because the underlying diagnostic information tends to be abstract and not readily describable in terms of well-defined image features. Representational Similarity Analysis (RSA) is a method originally developed in mathematical psychology that provides a theoretically sound and quantitative framework for measuring the mental representations of visual images in human observers. Here we used RSA to measure the extent to which the same underlying set of mammograms elicit similar mental representations in different practicing radiologists (N = 26). We found that the internal representations were statistically indistinguishable across different radiologists (p > 0.05). Moreover, the mental representations significantly parallel the diagnostic information in the images (p < 0.05 for each subject), indicating that various radiologists perceived the same set of diagnostic information in the underlying images. Together, these results indicate that medical images elicit similar mental representations in different radiologists.
UR - http://www.scopus.com/inward/record.url?scp=85080049011&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85080049011&partnerID=8YFLogxK
U2 - 10.2352/ISSN.2470-1173.2019.12.HVEI-225
DO - 10.2352/ISSN.2470-1173.2019.12.HVEI-225
M3 - Conference article
AN - SCOPUS:85080049011
SN - 2470-1173
VL - 2019
JO - IS and T International Symposium on Electronic Imaging Science and Technology
JF - IS and T International Symposium on Electronic Imaging Science and Technology
IS - 12
M1 - HVEI-225
T2 - 2019 Human Vision and Electronic Imaging Conference, HVEI 2019
Y2 - 13 January 2019 through 17 January 2019
ER -