
Episode 17:Impact of Medical Image Processing on Downstream AI Performance
Radiology AI Podcasts | RSNA
00:00
Is It Possible to Reduce the Number of Projections by 16 Times?
Lamia: Machines can see things that we cannot. And i think this is a good thing and a bad thing. The study by stamford and emory found that, as you mention, a i models can be trained to learn to detect race. Another study showed that deep learning models can identify gender from retinal fundus photographs. Neither one of these we can do, right? There's no way we il get a chest trate to day and can identify race. I don't think an authemologist can look at areginal fundis photograph and determine gender just based on those images. So for ai development, the implications are that you have to insure not
Transcript
Play full episode