Genealogies of Diagnostic Assemblages: Shaping a Radiological Sense

• In Event: Genealogies of the Human in AI Systems III.

Abstract

“Radiologists will lose their jobs to AI,” is by now commonplace in media reporting on the future of health care. In turn, STS researchers have urged to study automation practices not as replacement but as displacement, calling for a careful examination of new configurations of human labour and (automated) technical systems. This paper examines changing relationships between radiologists, engineers and computer scientists, focusing on how the work of the radiologist has been envisioned in relation to, as well as in competition with, emerging (automatization) technologies. Drawing on historical sources from the 1950s until the present, in the U.S. and Europe, we demonstrate how the core work of the radiologist was increasingly imagined as a combined visual detection and pattern recognition task. We claim that the emergence of new technical systems and concepts (such as the notion of “visual scanning” and “image interpreter systems”), tied to novel human-factors and psychophysiological research, helped shape a new conception of a radiological sense, that is, of the disciplined and distributed perception of radiologists and technologists. Tracing these genealogies of human-machine configurations in radiology (as diagnostic assemblages) helps us to understand current debates about AI and pattern recognition in image-based medicine based on shifting notions of “seeing” and “recognizing” patterns. This genealogical study is part of a bigger project (situated in the Netherlands) about AI in image-based clinical decision making.

Authors

• Flora Lysen, Maastricht University

• Sally Wyatt, Faculty of Arts and Social Sciences, Maastricht Univer