Let's Talk about Whole-Body Ultrasounds
Ultrasound is an amazing tech. It was a big part of my cancer diagnosis and we think of it as a little old-school but it’s cheap and extremely safe. Whole-body ultrasound in a scanner solves some real problems so this is super interesting and it’s hard to know where it leads.
I also have some frustrations with how MidJourney is pitching this!
Other folks will talk about whether this would just lead to a huge leap in incidental findings that will cause a lot of anxiety but not actually identify much serious illness. That’s outisde of my area of expertise except to say “I’m glad we’ll have more data to figure that stuff out.”
What I’m more interested in here is whether putting whole-body ultrasound up against MRI or CT makes sense and, look, it doesn’t. It doesn’t even make sense to put MRI up against CT. Not because one is better than the other, but because these scans do different things and give doctors different data, and if you don’t understand that you might go get a scan and think you’re all good when you only know a few things.
Elective MRI companies like to give you a xbig long list of the things that their scans can find, and the list is long! But they don’t tell you MRIs are bad at imaging parts of your body that move. This is we don't use MRI to screen for colon cancer or lung cancer (the colon moves more than you'd think). They’ll catch advanced cancers of those tissues, but these are two top killers and they’re bad at finding them early.
A chest CT is much more likely to catch an early lung cancer while colonoscopy can identify (and PREVENT) colon cancers (by removing pre-cancerous polyps.) Anyway, the point is…different scans do different things. So as we get excited about this tech, let’s talk about what ultrasound is good at and what it is bad at.
Ultrasound is really good at soft tissues, fluid-filled things, and things close enough to the surface that sound can get in and useful information can get back out.
This is why it’s great for looking at a thyroid, a breast lump, testicles, lymph nodes, gallbladders, kidneys, a fetus, blood flow, fluid around organs, and lots of abdominal stuff. It’s also why ultrasound was useful in my case. I had a lump. The lump was accessible. Ultrasound could say, “Yes, that is a lymph node, and yes, it looks weird.”
But ultrasounds have limitations that I would LOVE for AI to be able to overcome. It's just that there are some physics problems here...
The big problems for ultrasound are air and bone.
Air is a problem because sound waves do not pass smoothly from tissue into air. They mostly bounce. That's why the Midjourney scanner is showing a body being lowered into water...you can't do an ultra-sound through air. This is also why you get all jellied up when you go in for an ultrasound now. The scanner has to touch your body. But there's also air /inside/ your body. Your lungs are (unless something has gone very wrong) full of air. So ultrasound can tell you some very useful things about the surface of the lung and about problems that reach the surface of the lung, but it is not going to look through your lungs the way a CT scan can.
Could a very good AI do better with the messy information ultrasound gets from the chest? Sure. Better than doctors currently can? Maybe. Better enough to find small, early lung cancers deep inside aerated lung? I mean...show me the data and I'll believe it but, like, no.
AI can find patterns in data. It cannot recover information that physics prevented the scanner from collecting.
Bone is the other problem, which is also an issue in the chest but it matters even more for the brain because (unless something has gone very wrong) your brain is inside a skull. There are ultrasound techniques that work through the skull for specific purposes, but “whole brain scan that is basically like an MRI” is not a thing we should get excited about.
So when a company shows abdomen images and then says “like MRI,” I get twitchy. Abdomen is where ultrasound is strong, so focusing on that feels a little like them saying (yes, we know, this will be mostly good at the abdomen but the later in the same video being like "this will be like unlocking MRIs for the world!" which feels a little dishonest.)
Also, this person has SO LITTLE BODY FAT!! I am jealous. But also that can matter medically. Some bodies will be easier to scan than others.
This does not make the technology bad, but the pitch feels quite a bit broader than it should be.
A whole-body ultrasound scanner could be amazing for body composition, visceral fat, muscle mass, liver changes, kidneys, bladder, reproductive organs, some blood vessels, some lymph nodes, maybe some breast imaging, maybe things we have not figured out yet because no one has had cheap, repeatable scans at this scale.
And that's all potentially very exciting, especially because, for the beginning of this (if it catches on at all) there will be a bunch of rich people paying for that data to be collected and then researchers will be able to leverage that data to actually figure out things we can do with whole-body ultrasounds that maybe we otherwise wouldn't have guessed.
But will it be all the same things as we can do with a colonoscopy or an MRI? Of course not. An ultrasound will not be able to reach into your body and remove a polyp. MRIs are completely unaffected by bone.
Also, though, there is something exciting about a very cheap, easy scan like this: repeatability for an individual.
Medicine is often forced to compare “how you look today” with “how people usually look.” But a cheap, safe scan that people do over and over could compare you to you. You can look year to year and ask did something grow, or shrink, or shift?
If we really can do this cheaply and at scale, I think that could absolutely be an important part of the future of healthcare.
Or not! As much as it seems to a certain crowd like knowing everything about your body all of the time is the thing to do, most people don't act that way. But it could be important!
But “important” does not mean “the one thing we will do."
CT is great at lungs. MRI is great at many soft-tissue contrasts and the brain. Colonoscopy can see the inside of the colon and remove precancerous polyps while it’s there. Ultrasound is safe and cheap and real-time and good at lots of soft-tissue questions.
Different questions. Different techniques. Different answers. I just get the feeling both from the promo videos and the excitement generally that people are going to walk away from this thinking that it will be a one-stop thing.
If you walk away from this post with anything in your head let it be this...different scans do different things. We still have all the ones we have because they all fill in gaps that other scans don't have. Adding another full-body scan does not replace any of the others, but I very much hope that it adds.
