Errors in imaging reading and reporting

This paper explores medical imaging's epistemology, focusing on image reading challenges. It adopts "reliabilism," defining knowledge as reliably error-minimized outcomes. While AI lacks maturity, specialized physicians remain vital for image reading. Errors, up to 30%, persist in radiological reports. "Seeing-as," akin to Gestalt perception, further complicates, blurring error distinctions. Transparency and accountability challenges arise due to its complexity. The paper suggests avenues to reconcile knowledge accountability and the intricate "seeing-as" process, aiming to deepen the understanding of imaging's epistemology amidst these challenges.