Type:
Educational Exhibit
Keywords:
Education and training, Ethics, PET-CT, MR, CT, Artificial Intelligence
Authors:
T. Akinci D'Antonoli, T. J. Weikert, A. W. Sauter, G. Sommer, B. Stieltjes; Basel/CH
DOI:
10.26044/ecr2019/C-2553
Background
With the advent of AI,
we are facing a paradigm shift in our conceptualization of handling enormous amounts of data churned out with ever finer tools of reception and detection that are at our disposal in radiology.
Gone are the days when we,
as radiologists in the daily practice of our interpretation of cross-sectional images from patients,
were purely equipped with a general perception of the shades of gray or other color-coded pixels [1].
Thanks to AI,
we are now armed with the ability to acquire meaningful data that are otherwise beyond the range of our capacity of processing [1].
In radiology,
AI is about to enter the clinical workflows and has a wide range of applications: image analysis for lesion detection and characterization,
prognosis prediction,
computer-aided diagnosis,
workflow prioritization,
scheduling assistance,
image acquisition quality improvement [2].
Yet,
AI application in radiology is not without its problems.
It is imperative to understand the potential risks and problems that can arise with AI applications. This brings up the inevitable question: How should we govern AI from an ethical standpoint?