Medical imaging fails dark skin. Researchers fixed it.

EdenRester@kbin.social to News@lemmy.world – 114 points –
hub.jhu.edu

Traditional medical imaging works great for people with light skin but has trouble getting clear pictures from patients with darker skin. A Johns Hopkins University–led team found a way to deliver clear pictures of anyone's internal anatomy, no matter their skin tone.

In experiments the new imaging technique produced significantly sharper images for all people—and excelled with darker skin tones. It produced much clearer images of arteries running through the forearms of all participants, compared to standard imaging methods where it was nearly impossible to distinguish the arteries in darker-skinned individuals.

8

This situation reminds me of how they would only medically study male bodies and assumed everything would be applicable to women.

There's also a problem with facial recognition and darker skin pigment. I mean, why not throw a couple black people in the study groups!? Maybe then we wouldn't have all these issues...

The amount of melanin in one's skin is irrelevant in any sort of sense other than very real medical applications. Another area is tatoo removal lasers which won't harm white/lighter skin, but will damage people with darker skin pigment. It can cause both hyperpigmentaion and hypopigmentation as well as keloid scars in individuals with higher levels of melanin. (There are specialized tattoo removal lasers available for darker skin pigments.)

I remember a mini-controversy of home tattoo removal lasers (sounds like a bad idea in general to me!) that offered no warning to potential users that they could damage their skin depending on how dark their skin pigment is.

It just seems to me that there's an implicit discrimination in the populations used for studies. You'd think it would be obvious to try to represent each of the six skin shades on the Fitzpatrick scale with multiple participants if you're studying or developing technology related to our skin...

Edit: Smart watches that use lasers for the health measures are flawed as well. Again, that makes me think that people of color were clearly underrepresented in the studies developing this tech.

Appreciate this deep dive. Worth adding that there is a detectable bit of extra mortality and morbidity with fingertip pulse oximeters when testing darker-skinned folks. The devices usually face the same issue as smartwatches. This interferes with timely treatment for acute and chronic symptoms of COVID and other conditions which cause lowered blood oxygen concentrations.

Supposedly the tech was out there to solve it, but it was expensive enough that medtech manufacturers hadn’t ever bothered making compatible devices at a scale to serve most patients or consumers.

You're definitely correct about the pulse ox. That has also been known for a long time though. COVID just brought it to the public's attention.

I'm not sure that there is any alternative tech though. Pulse ox work by measuring the color of the blood at specific wavelengths.

I can't imagine that there'd be anything that non invasive around.

If you know of a citation, I'd be really interested.

Dr. Google yielded this good roundup of tech from around when I recall reading about it, near the end of the public health emergency. https://www.darkdaily.com/2023/05/22/innovators-develop-multi-analyte-pulse-oximeters-that-accurately-read-oxygen-levels-in-people-with-darker-skin-pigmentation/

Thanks!

Here is the most scientific version of this I could find from that article. I've got work to do, but I'll try to come back later to add something else.

There shouldn't be any barriers to entry for this. It won't be as cheap, but NIRS is not prohibitively expensive.

This is just really new. It takes lots of time for completely novel medical devices to enter the clinic.

This is also vastly more advanced than a pulse ox.

It was some sciencey site reporting on a journal article, I’ll see if I can find it again. Don’t remember the tech beyond a dim impression that it was “more emitters, different wavelengths, higher energy”

Tldr: this specific issue is more physics than racism. Plus there's more!

I presume op knows this, but others may want to look up WEIRD as an acronym.

It's actually both simpler and more complex than that, and it's less of an underrepresentation issue. It's not that that isn't an issue. Repesentation is just less of the cause of the problem in this context than you think.

It's also very much not explicit discrimination. The people who do this type of work are overwhelmingly anti racist. There's also plenty of dark skinned people amongst them. Most engineers are white males and Asians. That's an issue as well, but for another time.

First of all, they don't mention that this only applies to optical imaging. CT, ultrasound, MR, PET, etc. are not affected by skin tone (that's an ultrasound in the image, btw).

There are representation issues in medical testing. Not just purely racial ones as well. For example, most psychological studies are done with white women of college age. That's because the only people they can get to participate in such studies are psychology students who are overwhelmingly white women. Then there's the issue that you can't really experiment on children at all. That's WEIRD.

These are all well known and studied issues. They just don't have easy solutions.

The problem here isn't mostly representation. It's physics.

In order to image someone optically, you need to reflect light (or have something that emits light). Dark skin absorbs more and reflects less light than light skin. That is the purpose of dark skin.

No matter what we do, or how hard we try, optical imaging will never work as well on dark skinned people as light skinned people.

There's way more issues, too.

You can't just collect people based on skin color. Medical testing is extremely sensitive and difficult. You can't pay people too much. You can't pay people too little. Like the psychology issue, there are plenty of studies that are conducted on strange populations.

My brain is in several publicly available databases and papers and such. I get rooked into it because colleagues need subjects when I was in grad school. These studies are full of two kinds of people, grad students in the sciences, professors, their children, and people for whom a trivial amount of money is worth their time. That's not a cross section, and it's not really solvable.

You're correct that, in principle, you could collect people on quantitative skin color metrics for optical imaging. Other issues need a more complex understanding of race.

Now what is race? People say it's a social construct, but that's not scientifically correct. Race is a collection of phenotypes that helped our ancestors survive. It can be quantified. At the very least, genetically.

That's scientific racism. That's what the Nazis did. Granted, it's being done for anti racist purposes, but still, it's ugly.

So, I want to conduct a study that's DEI. How do I know I have proper representation? The terms Asian, black, Latino, and white are so huge and diverse that they are useless for this purpose.

What do I do?

Anyway, I could go on, but I'll stop after one last thing. The FDA is considering making DEI mandatory. That won't really solve the problem, though.

Edit: typos

A good example of why diversity in the workplace is a good thing.