AI algorithmically-enhanced motion suppression simulating an osteochondral defect in a young child

Abstract

Artificial intelligence (AI) in radiology has rapidly increased in our field and stands to allow more accurate diagnosis, quicker interpretations, easier workflows, and improved image quality. However, with superior image quality produced with the help of AI algorithms, one could begin to discount or even eliminate the review of nonalgorithmic enhanced images. At least currently, these images remain important. This case report demonstrates a unique anomaly simulating disease resulting from AI-enhanced motion suppression. On the original images, patient motion and an atypical linear motion artifact is obvious. However, the images reproduced using our AI motion artifact suppression algorithm suppressed nearly all (but not all) of the motion artifact resulting in what appeared to be an osteochondral lesion in a child's knee. This case illustrates the necessity for the interpreting radiologist to review both original acquisitions as well as AI-enhanced images, at least for the time being.

Keywords

• Denoising
• Artificial intelligence
• MRI