The Unobserved Anatomy: Negotiating the Plausibility of AI-Based Reconstructions of Missing Brain Structures in Clinical MRI Scans

In Antje Flüchter, Birte Förster, Britta Hochkirchen & Silke Schwandt (eds.), Plausibilisierung und Evidenz: Dynamiken und Praktiken von der Antike bis zur Gegenwart. Bielefeld University Press. pp. 169-192 (2023)
  Copy   BIBTEX

Abstract

Vast archives of fragmentary structural brain scans that are routinely acquired in medical clinics for diagnostic purposes have so far been considered to be unusable for neuroscientific research. Yet, recent studies have proposed that by deploying machine learning algorithms to fill in the missing anatomy, clinical scans could, in future, be used by researchers to gain new insights into various brain disorders. This chapter focuses on a study published in2019, whose authors developed a novel unsupervised machine learning algorithm for synthesising missing anatomy in extremely sparse clinical MRI scans of thousands of stroke patients. By approaching the study from a media-theoretical perspective, I analyse how its authors dis-cursively negotiated the anatomical and operative plausibility of the unobserved anatomy that their black-boxed algorithm reconstructed from the existing sparse data. My analysis fore-grounds the processual, relational, context-dependent, and essentially unstable character of the thus established plausibility of the algorithmically synthesised neuroanatomy.

Author's Profile

Paula Muhr
University of Zürich

Analytics

Added to PP
n/a

Downloads
201 (#85,410)

6 months
107 (#49,402)

Historical graph of downloads since first upload
This graph includes both downloads from PhilArchive and clicks on external links on PhilPapers.
How can I increase my downloads?