😲 It's The Oscar's For Radiology AI 2022 🥇

What the most influential Rad AI articles of the year?

Author

Mitch Mitchell
January 16, 2023

A Hoy Hoy,

I have the latest Rad AI info for you like Mr. Burns has money. Let's get into it.

Today's agenda:

  • The Imgy Awards!

    • Best articles of 2022 in Rad AI

Who will go home this year with the: Radiology Citation of Merit Award aka The Imgy!

Did you know that the Academy Awards Statues are nicknamed "The Oscar's" because of Margaret Herrick? Who's Margaret? She was just some academy member who said the statue looked like her uncle Oscar. And that's that - it stuck. It goes to show, crack more jokes because it might make you famous.

But Margaret isn't here to help us with our inaugural award ceremony, which will now be known as:

The Imgy's

The Imgy's looks at highly cited and influential Radiology publications that deal specifically with AI or related tech. Drum-roll, please.

Midjourney Prompt: An award statue, a gold statue of Wall-E, similar to the Oscar's statue. The statue is on a black base.

Best Visual Effects

//places award envelope on illuminator

The award for Best Visual Effects goes to: Quantum Iterative Reconstruction (QIR; Siemens Healthcare).

Cited 22 times in 2022, it was the 2nd most for all Radiology articles related to AI. This study showed how Quantum Iterative Reconstruction could improve contrast-to-noise ratio and make lesions easier to see. Is this AI? I think no, but it's an algorithmic method that uses stats and geometry to reduce image noise. It sounds like it might be the "same same, but different" of deep-learning reconstruction. Either way, with 22 citations I'm going to keep my eye on QIR.

The Altmetric Award

This award goes to the article with the most influence despite not necessarily being cited all that much. But don't worry, it's "influence" in the cringe way like the American Influence Awards <---a real thing 🤢.

The article focused on emerging AI technology in breast cancer detection. With over 280,000 new breast cancer cases in 2022, we will take all the help we can get.

Why are people excited? Well, the new tech has shown a lot of promise. They fed an algorithm >100k mammography images and asked the AI to classify each image with a score 1-10, with a 10 indicating the algorithm had the most confidence in cancer detection.

How did it do?

  • 86.8% of screen detected cancers were given a score of 10 (very high confidence by the AI)

  • 44.9% of interval cancers given a score of 10

  • 0.7% of screen detected cancers "missed" by the AI (given a score of 1).

No wonder this article is creating buzz. However, it may still be a while before this makes the jump from code-to-clinic. Congratulations to the AI evaluation of Mammography Exams article for winning this year's Altmetric Award!

The Imgy Award

This is the big one. What every author dreams about as a young child (despite the fact that this award ceremony is fake and didn't exist until just now): The Imgy Award of Merit.

This article lays out the path that I think holds the most promise for AI in Radiology and healthcare in general. That is one of augmentation. Using computer-aided diagnosis is like stepping on a star in Mario Brothers. A massive power-up.

Power-up Moonwalkin'

By using computer-aided diagnosis, readers improved at interpreting chest CT scans with indeterminate pulmonary nodules. Is there any reason we wouldn't want to introduce technology that improves performance? Merely assisting physicians with technological aids will be a very important step in using AI in the clinical setting. I am excited to see more studies of AI that seek to improve physicians instead of replacing.

I'm getting the "Wrap it Up" music so that will be all for today. I'll see you next week for more With and Without AI.

-Mitch

If you liked this piece please subscribe below: