4 Wege, wie künstliche Intelligenz die Medizin verändern wird – KI kann Tausende von Bildern vergleichen, um gefährliche Muster aufzudecken, Scans mit ultrahoher Auflösung aus Bildern mit niedriger Auflösung erstellen und sehen, was das menschliche Auge übersieht.
https://www.ucsf.edu/news/2024/12/429031/4-ways-artificial-intelligence-poised-transform-medicine
4 Comments
From the article
>The radiologist was dead.
>Or at least that’s what artificial intelligence (AI) experts prophesized in 2016 when they said AI would outperform radiologists within the decade.
>Today, AI isn’t replacing imaging specialists, but its use is leading health care providers to reimagine the field. That’s why UC San Francisco was among the first U.S. universities to combine AI and machine learning with medical imaging in research and education by opening its [Center for Intelligent Imaging](https://intelligentimaging.ucsf.edu/index.php/).
>Take a look at how UCSF researchers are pioneering human-centered AI solutions to some of medicine’s biggest challenges.
of all the medical uses, upscaling is not a good one. The last place you want AI hallucinations is in diagnostic imaging.
Oh great, one to diagnose you and another to deny you care.
Reading “create ultra-high resolution scans from low-res images” is hilarious for how bad of an idea that is. Nothing, AI or otherwise, can just manifest additional information that wasn’t present in a lower-resolution image. Anything you see in an AI-upscaled image that you couldn’t make out in the original lower-resolution image is just something the AI created, and there is a very high probability that it’s just a complete hallucination. It’s not _impossible_ that an AI could pick up on subtle clues that we cannot, but you’d still inevitably get far more incorrect results than correct ones.