Midjourneys New Scanner Could Change Everything

TITLE: Midjourney’s New Scanner Could Change Everything CHANNEL: Dr. Know-it-all Knows it all DATE: 2026-06-18 ---TRANSCRIPT--- So, Midjourney just announced a full-body medical scanner. Yeah, Midjourney, the company best known for generating images with artificial intelligence, now says it plans to build a global network of scanners capable of performing 60-second full-body scans using ultrasound technology, with the first location expected to open in San Francisco in 2027. At first glance, this seems completely disconnected from what Midjourney actually does. If someone had asked me yesterday which AI company was most likely to enter healthcare, I probably would have guessed OpenAI, which actually just did that, Google, Anthropic, or maybe even Tesla. Midjourney, on the other hand, would have been somewhere near the bottom of the list. But the more I looked into this announcement, the more I started to think there may be a deeper connection. Because Midjourney has never really been in the business of making pretty pictures. What the company actually does is reconstruct reality from incomplete information. Before Midjourney founder David Holz created Leap Motion, which reconstructed the position of your hands from sensor data. Midjourney reconstructs images from noise, and now this new company is attempting to reconstruct the interior of the human body from ultrasound signals. In other words, all three projects are variations on the same fundamental problem. The really ambitious part is not just the scanner itself. The scanner is just the hardware. The much bigger idea is that if you make medical imaging cheap enough, fast enough, and accessible enough, you can start collecting body scans at a scale that has never existed before. Instead of occasional MRIs ordered after a medical problem appears, you could potentially have millions of people generating continuous records of their health over years or decades. And that leads directly to the question that makes this announcement so fascinating to me. Can artificial intelligence compensate for the limitations of cheaper hardware? Because MRI machines are extraordinarily expensive and extraordinarily capable. Ultrasound systems, on the other hand, are much cheaper, but also much less precise. Midjourney’s apparent bet is that enough data, enough compute, and enough AI reconstruction can eventually close that gap and perhaps even surpass it. If they’re right, the implications are enormous. If they’re wrong, they will discover that healthcare is one of the hardest industries on Earth to disrupt because success requires much more than good engineering. You have to overcome the limits of physics, survive years of regulatory scrutiny, prove the system actually improves patient outcomes, and very importantly, avoid creating a false positive nightmare that overwhelms doctors and patients alike. There is a tremendous amount to unpack here because buried underneath what sounds like a bizarre product announcement might be one of the most ambitious attempts yet to use AI to fundamentally change how human beings understand their own health. Let’s take a look. Before we start, a quick shout-out to my channel sponsor Joa. They make amazing accessories for your Tesla and other EVs and also have great warranties and customer service, too. In fact, I use their accessories daily. Be sure to check the link in the description to get 5% off a fan-cooled phone charger, a portable tire inflator, a fold-out lap table, and a lot of other items. Oh, and they make perfect gifts for you and your EV-loving friends, too. So, be sure to check out Joa, and now let’s get back to it. Hey, y’all. It’s Dr. Know-It-All. I woke up this morning to quite the flurry of activity about Midjourney. I had, as usual, another video that was planned and halfway edited and stuff, but this was important enough that I thought I would supersede that video and show this instead. So, stay tuned for that other one. Make sure you’re subscribed and hit that bell notification icon so you catch all my videos. But anyway, let’s talk about Midjourney Medical and their full body scanner announcement. I’m going to kind of zip through this intro video. Of course, I will leave links to everything in the description so you can look at them at your own leisure. But anyway, you can see here that they have this product called Midjourney Medical Full Body Scanner that you go ahead and stand in. This is sort of like a bathtub, basically. So, you can see the person gets in this thing and then they are lowered into this tub of water here. So, lower, lower, lower. And then once she’s inside the tub, the scan actually begins and you can see it takes about a minute to do that. And then it does a full body scan, unfortunately only up to your neck. It does not do your head, at least at this point, because you’re obviously going to be a little uncomfortable having to hold your breath for a full minute. But anyway, the whole thing apparently happens in only a minute after you’re submerged up to your neck in the water. You then are lifted back out again, and the scan is completed. I do want to show this little bit here. So, this is obviously a prototype device here. You can literally see C-clamps that are holding this whole thing together, bunch of bare wires and things like that. So, while this is a cool demo, this is clearly a prototype prototype device here. The real device will probably be assembled in a factory and be able to ship everywhere as a kind of a unit that you install. The whole point of what they’re trying to do here is use relatively inexpensive ultrasound technology and mass produce that to save a lot of money, and then install thousands and then hundreds of thousands of potentially millions of these devices around the world, which will allow people to go in on a regular basis, perhaps every year, perhaps every 6 months, perhaps every 3 months to get a full body scan very, very quickly in a kind of a spa-like experience. You can see that the lighting is very soft and everything. The whole point is it’s very relaxing. My wife has terrible claustrophobia, and the couple of times that she’s had to have an MRI, I’ve had to go in with her. She’s had to have Xanax. I’ve had to pet her head or pat her arm or something like that to keep her from freaking out. This sort of scan would be much more pleasant than that. It would just sort of like be taking a bath or something like that. So, the water is at body temperature or a little bit warmer. You feel very relaxed. You just stand there for a minute, and you come back out, and you feel like you’ve been in a spa rather than in some sort of claustrophobia-inducing tube. And Robert Scoble, who lives in the Bay Area, lucky guy, because he gets to go to a lot of these events, said, “I don’t often walk out of a product introduction thinking, ‘Wow, I just saw the world change for the better.’” And apparently got one of the few pictures of David Holz at this event, because apparently the art museum where it was held in does not allow photography, and and so they told everybody to get rid of their phones. And if you think this isn’t a big deal, Robert also posted, “My best friend Brandon Wurtz, a brilliant AI pioneer, died at the age of 40 from colon cancer. The doctors miss it, this would catch it. And that’s really what this comes down to. An MRI is so expensive and there are so few of the devices in the world that you have to have some sort of medical urgency to get one of the scans. The whole point of this is that you could do this on a very regular basis and actually catch things much, much earlier because you could catch body changes prior to any symptoms developing and therefore have a much higher chance of a successful outcome than waiting until symptoms appear, at which point it might be too late. And then Mark Kuchman has a really nice summary of this. This new Midjourney medical scanner project is wild in the best possible way. A 60-second full body scan using ultrasound instead of radiation or giant MRI magnets wrapped into something that feels more like a spa visit than a hospital appointment. The first location is planned for San Francisco in 2027 as we noted. The real ambition is much bigger though. A whole network of these scanners eventually scaling to tens of thousands of machines and regular body scans for huge numbers of people. Of course, the hard part is not the announcement, it’s the regulation, validation, accuracy, false positives, cost, and proving this actually works in the real world. But as a direction, it’s very exciting. Cheaper, more frequent, less invasive health data. And then of course, we have a series of pictures of what this would actually look like as you did the scan. All right, so what does this thing actually do? Well, it has a ring of 40-plus ultrasound on-chip sensors. You stand in water again up to your neck. It can’t do your head, at least not at this point. The scanner then, similar to an MRI but using very different technology, an MRI actually uses magnetic resonance of the protons in your body. It puts you in a gigantic magnetic field that aligns the protons in the same direction and then it pulses those. And the way that they oscillate back again gives a bunch of information. It’s absolutely fascinating technology, but it requires superconducting magnets, helium baths, a huge amount of electricity, a massive amount of magnetism, and an extremely expensive device. This on the other hand uses sound in water. It actually has tiny little, you can imagine as speakers. They oscillate quickly back and forth and the water transmits that information to the person’s body. If you or someone you know has been pregnant and you’ve seen the ultrasounds, that’s exactly what this is. It’s just on a full body scale rather than a concentrated focused ultrasound beam that’s just for a small area of the body. So anyway, the scanner using this ultrasound captures vertical slices. The whole thing is done in 60 seconds, which is pretty crazy. Apparently, the scanner has two petaflops of compute, which is absolutely crazy and it produces 3D body composition maps like you’re seeing in the images. So the deal here is this is very different technology than MRI. Ultrasound is amazing, but it’s much less precise than MRI and at present it can only reveal a certain amount of information about the body. The goal of Midjourney Medical is to make this scanner as precise as MRI, but using much, much less expensive technology. And so like I said in the opener, what they’re doing is using very low resolution information, very noisy information, the stuff that’s coming from ultrasound and using AI to reconstruct something that’s as good as an MRI scan or potentially better than that. That’s what they say. So this is a fundamentally AI native approach to imaging and actually interestingly enough, there’s kind of an analogy with Tesla using vision only full self-driving versus other companies that use lidar. Those companies have lasers that scan the environment. They shoot out beams that discover what’s around them in the environment, whereas what Tesla does is it uses cameras plus a bunch of AI compute to reconstruct that three-dimensional world around the Tesla with much, much cheaper hardware. Midjourney is using the same basic concept here. They’re using cheaper technology, cheaper hardware that’s much more available and then using compute and AI to make up that difference to create images that are as precise as MRI images. And with that, I bet you’re starting to see why this makes sense for Midjourney to do this. What the Midjourney imaging and video product does right now is recover reality from incomplete information. You give it a prompt and it starts with noise and it creates something that’s highly believable. Well, if you start with a noisy physical image of an actual human body, the goal here is to recreate the reality of what that would look like if it was denoised. And that’s the reason why it actually makes sense for Midjourney to do this when you stop to think about it. So, the reason this really matters is you were replacing very expensive, extremely precise hardware and proprietary software with tons of data, tons of compute, and more general-purpose software. So, with an MRI, you look for bigger magnets, better detectors, more expensive machines, and of course better software as well because the software has to reconstruct what it’s seeing. But, with this new AI model, what you’re doing is replacing that very expensive machinery with more data and better AI models to create a cheaper sensor suite plus enormous data sets plus AI reconstruction that can eventually perform the same as or potentially even outperform very expensive MRI machines. And to be frank, ultrasound is already moving into this space. The FDA recently cleared ultrasound AI systems that are trained on millions of scans. We have AI-driven prediction systems that use standard ultrasound images, and we have AI-enhanced 3D ultrasound reconstruction that is in research right now. So, this idea is not completely new. What is new is the idea of doing a complete body scan at a reasonable price and collecting a massive data set that will allow this reconstruction to get better and better and also of course and most importantly to get better and better at discovering any anomalies. In other words, stuff that you need to take a second look at because something looks like it could be wrong. So, this all sounds fantastic, but there are three basic issues with this that could stall progress on the Midjourney medical scanner. The first one is just a basic physics problem. MRI is really good. It provides excellent soft tissue contrast. It provides deep penetration throughout the entire body, and it provides highly detailed imaging. For ultrasound on the other hand, air blocks the sound, but obviously that’s why the person is in the water because that provides better contact. Also, bone block sound, you have to compensate for those shadows. You also have variable penetration and of course you have lower resolution. The resolution is just not as good as MRIs out of the gate. So, first and foremost, Midjourney has to fight the physics of the fact that an ultrasound is just not as good as an MRI by default, but that of course is where their massive AI experience comes into play and to my mind that challenge is likely already taken care of. They’ve been working on this for over a year in secret and likely they feel like they have a really good trajectory to get to solving those physics problems. The next problem is the false positive trap. Suppose you have 100 million healthy people scanned annually. Within that data set, you’re going to find harmless cysts, benign lesions, anatomical weirdnesses, all sorts of things that might look scary, but they’re completely normal. The problem would be if the scanner identified a significant portion of these as potentially problematic things because then while the scanner could technically be brilliant and work amazingly, it could still create more harm than good if it generates huge numbers of unnecessary follow-up procedures, huge numbers of follow-up MRIs, and even potentially surgery that is unnecessary. Those would be terrible outcomes and you really don’t want a lot of false positives. And if you think about it, medicine really isn’t defined by its technology, it’s defined by its outcomes. What you want is to discover a problem, it doesn’t really matter how you discover it. You want to discover that problem early on and take care of it. You also don’t want to think you have a problem when you don’t because that will waste a whole bunch of people’s time and money, but also create the boy who cried wolf scenario where if you have four or five false positives, you might not go to the doctor the one time you actually needed to. So, getting false positives down is going to be a huge challenge for Midjourney Medical. And then finally, as you might expect, is the regulatory hurdle. So, if you listen to the introduction carefully, Midjourney Medical appears to be launching itself initially as a wellness, body composition, health tracking type of device, not a diagnostic device. So, why would they do this? Number one, they need more data. They need to collect data from people in order to get better at doing what they’re doing. But, then second of all, of course, is the regulatory hurdles. The FDA, the clinical trials, the validation, the sensitivity, the specificity, and of course, the medical liability. If you give false positives, or even worse, if you give a false negative, if you say you came out fantastic, there is nothing wrong with you, and then it turns out there was a problem, that could hold the company liable and ultimately bankrupt it. So, if we’re going to move from wellness and body composition promises to actual diagnoses, that’s going to take years of data and funding to get approval. So, while the scanner is a physics and engineering problem, the diagnosis itself is a regulatory challenge, and that really matters. All right, so that’s the skeptics’ take, but what happens if this actually works? This could be civilization-changing. Just imagine if the cost of this, especially with an insurance copay or something like that, could approach the cost of an annual physical or routine health check, and just as importantly be available, you know, at the corner CVS or something like that, rather than having to schedule an expensive MRI at only one or two locations per city, and have a specialist referral to even get that. That changes things significantly. Humanity in that case would move from reactive medicine to longitudinal proactive medicine. In other words, you could stop comparing yourself to averages and wait until something symptomatic appears to comparing yourself to yourself over long periods of time, and potentially discovering problems long before they become symptomatic. Those are life-changing both for individuals and for civilization as a whole. So, let’s end up with a little bit of speculation about the probability that this is going to happen, and the potential for it to scale out to what Midjourney wants. So, will the first scanner appear in San Francisco in 2027? I give it 80 plus percent likelihood. I think it is very likely that that will happen. Midjourney is [clears throat] quite clearly serious about it. They’ve already got a prototype device built. They’re recruiting people to work on this. That is highly likely to happen. Will the wellness version of this product succeed? Yeah, I give it about a 60 or 70% chance for that as well because people already spend absurd amounts of money on wellness and optimizing their longevity outside of specific medical diagnosis. So, people will spend money on this. It will very likely succeed as a wellness product. Will it get to MRI comparable imaging? I would give that about 30% odds within the next 5 to 10 years. It’s an extremely difficult physics and engineering problem to get something that’s that noisy and that low resolution up to the resolution of an MRI machine. So, I give it a bit less than 50% chance, but I think if anybody could do it, Midjourney really has the chops to do that. Next up, will it become an FDA cleared diagnostic platform? Maybe 20 to 25% likelihood. It’s really connected to how likely it is that they’re going to get to MRI comparable imaging. So, the two of those are tied together, although obviously have to get to the quality before you can get to FDA clearance. So, that’s why I give FDA clearance a little bit lower likelihood than MRI comparable imaging. And finally, will it do what they claim and have civilization level impact? That’s probably on the order of 10 to 15%. And that sounds really low, but the catch is if this actually works, the impact is enormous. The impact of it working is so outsized that a 10% chance is really worth it. In the end, this is one of those rare projects where failure means almost nothing changes, but success fundamentally changes how humanity understands its own health, and that’s exactly why it’s worth paying attention to and exactly why Midjourney is doing this and exactly why we should all applaud them for trying such an ambitious project. All righty, folks, that’s what I’ve got for you today. Let me know in the comments what you think about all of this. While you’re down there, if you don’t mind liking the video, it really helps other people to find it. And if you want to help out the channel the most in 2026, help us get to 100,000 subscribers. We’re getting real close. Please do consider subscribing, hitting that bell notification icon, and I will see you in the next video. [music] Bye-bye.

[music]