Artificial intelligence for abnormality detection in high volume neuroimaging: a systematic review and meta-analysis

Siddharth Agarwal,David A. Wood,Mariusz Grzeda,Chandhini Suresh,Munaib Din,James Cole,Marc Modat,Thomas C Booth
2024-05-09
Abstract:Purpose: Most studies evaluating artificial intelligence (AI) models that detect abnormalities in neuroimaging are either tested on unrepresentative patient cohorts or are insufficiently well-validated, leading to poor generalisability to real-world tasks. The aim was to determine the diagnostic test accuracy and summarise the evidence supporting the use of AI models performing first-line, high-volume neuroimaging tasks.
Image and Video Processing,Computer Vision and Pattern Recognition
What problem does this paper attempt to address?