Midjourney Medical(midjourney.com) |
Midjourney Medical(midjourney.com) |
Maybe I'd even underpay a few people in developing countries with experience reading ultrasounds to check over the images so that if the humans detected anything suspicious I could give my sucker/client something more specific to tell their doctor about. That'd probably get me some good PR on social media as people post about how my fancy spa found their massive tumor or whatever.
Then I'd use their body scans as training data for my image generating AI. The waivers I'd have people sign to use the service would make sure that I wasn't at risk of any thorny legal issues from the use of all those images for training unlike the rampant copyright infringement method I'd been using previously and would also make sure I couldn't be held responsible for anything my scans found or didn't find.
Less cynically, maybe this thing will be nothing at all like that and one day it'll end up being used by real doctors in actual hospitals and save a bunch of lives or something. Who knows.
The other part wonders if this is the next clinkle.
MJ has shipped stuff before though so I’m optimistic.
But hey if not, actually cool.
But, even granting they could be true, they would be true under the status quo.
Sure, a one off full body scan might be scary and lead to unnecessary action. But if a technology of the sort being described here were to exist, you would just get daily (or more frequent) scans to monitor the situation. Is that tumor actually growing or is it just a transient thing your immune system is dealing with? Way easier to tell if imaging is cheap, fast, and frequent.
And then there is the data.
No one knows what is actually going on in our bodies. If we had the ability to do billions of scans, imagine the longitudinal studies that could be performed.
It would radically alter medicine.
- patients will worry too much, and - it will cost time and money to investigate.
Both spurious rationales cooked up by an industry that is at least as hostile to humanity as it is helpful.
US is a good diagnostic tool, but it can be challenging to read because obtaining good images is very operator dependent. You need to have a good sonographer that can get the right views, knows how to adjust the imaging parameters to produce high quality images. It's not like CT or MR where the tech just sets a few basic scanning parameters and let the machine do its job.
However, see my other comment, the example images they provide on the page do not look great, very limited organ detail.
edit: clarification
Chest cavity, brain tissue scanning etc. will likely remain unrealistic as ultrasound waves won't penetrate bone and the ribs and skull will interfere.
So im curious to know the limitations of this device
The images and description of the launch seem like they are behind where my buddy was 10+ years ago - so I expect a pretty difficult road ahead, between getting to where it's actually medically viable, and then stomaching the FDA process.
my second reaction: maybe it does? did they hire up an army of physicists to make better diffusion models or something and they actually have people on staff who can do this?
Congrats!
...what. You descend into water and it scans your whole body? How do you breathe? How do you come out the other end?
Have they actually invented some type of novel scanning technology, or is this just AI slop gone wild?
If it is then wow!
Remind me of this, radar based.
It looks like a legit attempt. Wow. This is insanely innovative.
Medical imaging is literally the last of the last places where you want to hallucinate a tiny little blob.
Only after that happens will I have to even consider how comfortable I am with the idea of handing over what they suggest will be massive amounts of highly personal medical data to this company and how much I trust them not to exploit that information for their own purposes and profit.
Medical I don't care about futuristic sounding stuff. Just show me evidence based and clinically useful testing.
Use AI and new scans to help sure but prove it works otherwise this could be another dead end.
Outside of providing access to their core AI products at a free or discounted rate, what philanthropic initiatives are OpenAI and Anthropic pursuing to improve the lives of people in developing countries?. I can't recall seeing anything on their blog recently about it. Happy to be corrected.
The spa approach is a little weird. FDA workaround?
So they get more data of the same person over time.
If every hospital had one, even if they sat idle 90% of the day, thats enough to hit that target.
For what possible reasons? Are people going to be doing these things recreationally? Cause otherwise you're talking about scanning the entire world's population, including the very young, the very old, the mobility-impaired, and those without easy access to US-based facilities (i.e.... people who are part of the small fraction of the global population who do not live in the US), twice over, every 18 months.
What possible use could there be for doing this?
I recognize that the presser says the scanners will be deployed "around the world," but let's be real, this will probably be 80% US.
Sounds good to me.
Not a physician, I wonder about the general efficacy of random scans vs more boring traditional things like bloodwork. That is: is there more clinical power in doing blood + urine labs monthly or body scans like this?
> Our ambitious goal is by 2031 to have a fleet of over 50,000 scanners worldwide - with a total scanning capacity of a billion scans a month - enough to cover a huge percentage of the global population, or enough to give regular, monthly scans to a billion people.
> What This Leads To
> Whether or not our scanners are a service that everyone uses, to us, the most important thing is that everyone will be able to use them.
There is no way these will be available to a billion people. This is a luxury product for rich people, which is fine, but they cannot afford to run these for a billion people every month. Think of the infrastructure—both human and physical—to provide that. Think of the distribution of wealth across the world. Come on.
There are so many small, boring details that will have to be ironed out: many Americans won't fit in that machine, kids will not sit still, you'll have to clean them constantly (people pee in warm water), buying and re-tooling property for spas with zoning and licenses is arduous and jurisdiction-specific, etc. etc. etc.
What they are pitching and focused on (data, models, tech) is the fun part. It's not nearly most of the problem.
I'm not sure if they believe this (naïve, unserious) or if they don't (lying). Either way doesn't build trust.
Where's your sense of fun and adventure? /s
I've encountered this attitude before, and I always find it perplexing that there are people who are annoyed by, even hostile to, the idea of frequent health telemetry.
What possible use? How about giving people greater visibility inside their own bodies without having to navigate the labyrinth of the healthcare machine and without having to justify themselves to actuaries?
There’s a reason most people don’t get medical scans every checkup, they’re simply not necessary for the majority of (healthy) people.
Surely, whatever this is giving you, getting a scan once a month must be plenty. They need a billion people to get a scan every month.
My wife's a cardiologist and hypochondriacs with smartwatches have become a frequent occurrence because healthy young people despite regular check ups have convinced themselves their watch telling them their pulse got high that one time must mean they're dying and they'll show up not one but five times.
The same is happening with so called "sleep optimizations" which themselves can produce insomnia as people start to self-monitor and enact sleep efforts.
The point is to generate an enormous unlabeled dataset. Historically, ML for medical imaging depended on a small number of labeled images - small because you needed to have an expert study the image and label it as healthy/cancer/etc. But the "GPT breakthrough" was that it was better to use vast unlabeled datasets - in the case of LLMs, text - than small labeled ones.
I think this is currently seen as too expensive to do for people who have lower risk, but I mention it as an example of something that one could check for more routinely given much cheaper ultrasound scans.
Prophylactic ultrasound exams are also apparently much more plausible on medical cost/benefit than prophylactic CT exams, because the CT exams very slightly increase one's cancer risk (https://xkcd.com/radiation/), where ultrasound doesn't.
(At a friend's doctor's suggestion, I started taking alkali citrate supplements and switched from almond milk to oat milk; I now apparently rarely get kidney stones.)
Hopefully it doesn’t become Gattaca.
Early detection of disease, as well as every kind of physical issue with the body you can imagine.
The incredulity of the question seems rooted in the medical culture of our current time. It's easy to imagine a science fiction future where scans happen not every 9 months, but daily, in your home, and the idea of not constantly checking your full body would be as strange as not brushing your teeth is to us...
Umm...the same use we get out of an annual physical or dental checkup.
We are well on our way to that classic scifi trope of the villian being introduced as they soak a special tub of goop. (Dune, GOTG, Star Wars)
That's about 1 scan per unit, every 2 minutes, 24/7.
To the extent you can really call pointing their behaviour out as victimizing them, I would consider bad PR to be a fair tradeoff.
The whole argument that "you'll worry yourself sick" is such patronizing trash. It's obviously programming that came from the insurance industry, and you lapped it right up.
You're not arguing in good faith when you equate those.
> The Ningen Dock is a comprehensive health checkup system that includes a battery of tests, including blood tests, chest X-rays, and ultrasound scans, among others as well as advanced diagnostic tests as Magnetic Resonance Imaging (MRI), Computerized Tomography (CT) or Endoscopy. These tests can help detect potential health problems early before they become more serious or difficult to treat.
The fact that doctors like your wife think that people who are concerned about their health and want more information is a problem tells me everything I need to know about your (and her) worldview. You've dressed it up as being pragmatic, but the reality is that you're arguing for censorship and against freedom of information.
It is a problem because there's evidence based standards for when examinations are indicated and prolong or improve a person's life. You being extremely concerned doesn't move that needle and subjecting you to tests simply because you're anxious is blatantly unethical and harmful to your psychological wellbeing.
And nope this isn't censorship, it's being mathematically literate and understanding how data production works. Here's an actual real world example. There are aids tests that are 99% accurate. About 30 in 1000 people in the US have AIDS. 99/1 is great odds, let's test everyone, data doesn't hurt right? Except as it turns out if you test a thousand people randomly you'll have 10 false positives and 3 people with AIDS, Bayes in action.
So if you sent every American through body scanners, which are less reliable than that test btw, you'd have quite literally millions of people in follow up procedures for diseases they do not have with their mental health ruined and the system ground to a halt, because producing information is not always the right thing to do.
Maybe your employer pays for you to get a more comprehensive checkup by default and you're unaware of this?
But the ones vast majority of population here gets do not include MRI or CT or Endoscopy.
And, _even then_; specific checkups when you're looking for _specific things_ are still very different things than a full-body MRIs.
Instead of casting a net of unknown quality every month, comparing against a null dataset (there does not exist a large dataset of these scans with outcomes for given markers).
Why not advocate cheap, easy blood/urine tests with higher frequency? Those tests do have large reference datasets with outcomes. And they have prescriptive value: there is likely more benefit to catching hypertension or diabetes earlier in more people.
Going forward into the future and not measuring more accurately because we are worried about false positives in our current limited understanding is a very conservative take.
On what basis do you say this? There is an extensive literature that refutes this. Scanners have been getting much better since the first CT scans and many more people are getting them.
I mean, with that much data, you may be able to understand under what timeframe a tumor is actually of concern. What's so bad about having some false positives?
Having invasive surgery. Undergoing chemotherapy. The former is bad, the latter is basically a 'lets hope it kills the cancer before it kills you' situation.
It's arguable which one is worse, but I'd rather not have to ever partake in either of them again.
Given the source, I will treat it as nonsense science fiction until it’s built, functional and scientifically tested.
I'm also following the very inspirational journey of the former Gitlab CEO who battles cancer by founding companies with his own money [0].
FUCT, huh? Genius marketing move.
What the hell are they talking about. This is no way real and a late April fools joke right? Right?
This is what came to my mind first too. It feels like the sort of thing you could come up with after a lot of ‘that’s a great insight!’, with the LLM eventually projecting absolute certainty that it’s a ground-breaking idea that’s definitely going to work.
I’m not sure whether I like that this is my knee-jerk reaction.
Do they have any sort of prototypes of this hardware that’s going to be working reliably in their custom-built spa in the notoriously difficult-to-get-permits-in San Francisco by the end of next year?…
I guess they pivoted from making ai-artwork to ultrasounds?
What do you mean here?
The idea came from LLMs? They built this with LLMs?
This tends to create a feedback loop where unsound ideas are amplified.
2,592,000 seconds / 20,000 scans = 129.6 seconds/scan
If you really hate your customers and don't care about cleaning out the tanks between scans, you could make this work. They have to be either able bodied to be able to move in and out quickly enough, or if they're not you just toss them unceremoniously onto the platform and drag them off after.
Realistically, a 60 second scan is going to take ~10 mins minimum, and will operate 8 hours a day, let's say charitably 7 days a week. Assume 50% utilisation due to staffing, repair, holidays, etc, we're looking at ~36m a month, or 0.036% of what is being pitched here. (8hrs * 6 scans * 30 days * 0.5 utilisation * 50k machines).
> That, collectively, we can begin to change our relationship with our bodies and start to ask questions like: if we can catch things early, can we change our lifestyles to correct them?
We can already ask this question...
> And seeing our bodies change over time, alongside our actions, how much can we improve our health, our minds, and our lives?
Again, we can already ask this question
> We think it's completely possible that with enough early imaging in the future, the world could avoid 30% of all deaths and 50% of all healthcare costs. The cultural, physical, and mental health benefits of all of this are hard to comprehend, but also hard to overstate.
What? I have no idea what is meant here by "hard to overstate".
> You want as much data as you can get about your health as quickly and as cheaply as possible. In other words, you want a technology optimized for getting as many “megabytes per second per dollar” of information about your body.
Thanks for including the "megabytes per second per dollar" unit breakdown, I didn't understand the first sentence at all without that!
> And we live longer, healthier lives, better lives.
More AI slop
> When you step into the water, you’re standing on top of a platform. The platform is connected to rails and begins to descend into the water - an elevator gently lowering you at around 2 inches, or 5 centimeters, per second.
More AI slop. You'd only be done in 60 seconds if you're exactly 5 feet tall
I guess some type of software platform would add some competitive distancing?
I get the benefits of regular scans although I also know that they tend to catch a lot of otherwise benign tumors that can cause a lot of stress.
I also found this researcher on their staff who studies tomography https://scholar.google.com/citations?hl=en&user=idvD2yYAAAAJ
With a big enough data set of [all kinds of bio values, including ones considered irrelevant for that disease] labeled with diagnoses, I suspect we could get very fast and accurate automatic diagnoses, even from a limited data set currently considered uncorrelated. Rather than going to your primary care physician, you'd go into the standardized, mass-produced and thus reasonably cheap everything-scanner, and you could likely get a more accurate diagnosis (or at least "things to check") than the average doctor would be able to give you under the practical constraints they typically operate under (time, available information/diagnostics).
This goes in that direction, and I'm really excited to see where it goes. I could imagine that given enough training data, ML models will be able to pick up on minute details that make it possible to diagnose diseases that weren't historically considered ultrasound-diagnoseable from this kind of detailed ultrasound.
I think combining it with gas chromatography/mass spectrometry of e.g. breath or blood/sweat/urine samples would also have the potential to be a cost-effective diagnosis method - lots of data, probably not all too useful for human interpretation, but would open the potential to walk up to a machine, breathe into it, spit into it, pee into it, give it a swab, and have it come up with an accurate diagnosis without invasive testing. If mass produced, the cost of something like this could easily drop below the cost of a typical doctor's visit. (I googled it and it seems like GCMS is already used for some diagnoses, but screening only for a few specific diseases rather than "throw ML at it and try to diagnose everything").
While there are many individual stories of full-body scans detecting early-stage cancer before it became symptomatic, there seems to be a general sense among doctors that implementing full-body scanning on a population level would lead to overall more harm than good. The thinking is that it is better to do regular targeted screenings for diseases that you're in a risk group for (e.g. colonoscopies, mammograms, cancer marker blood tests, etc.) rather than full-body scans.
I'm not a doctor, and I personally do find the idea of full-body scans very appealing, but I also know that if the scan detects a possible cancer, I wouldn't be able to just ignore it if the doctor tells me it's likely ok. Any time I felt any pain or any sort of symptom in that general area, I know I would worry about it. Maybe that's worth it for the potential life-saving results, but it definitely is a cost of this type of scan that needs to be acknowledged.
But we would have great data over time, both individually (weird tends to only matter if they are changing) and as a population.
It is neither controversial nor complicated to detect some cancers by scent.
Taking the "headspace" of something is also not really complicated.
There are people who can reliably smell/detect Parkinson:
https://www.npr.org/sections/health-shots/2020/03/23/8202745...
It might also reveal that every MRI shows ghost artifacts a half a dozen times that make it longitudinally useless, of course. I'm not foolish enough to think that epidemiologists haven't thought of this.
There's a reason why billionaires like David Rockefeller, Larry Ellison, and Rupert Murdoch are able to live much longer lives than average, and having an oncall health team (that I'm sure does frequent testing and monitoring) is a big contributor to that.
More testing and data collection doesn't mean that every single anomaly would need to be investigated or communicated with the patient, but would provide a better longitudinal view that can help with disease prevention and health optimization.
Bayes Theorem: https://en.wikipedia.org/wiki/Bayes'_theorem
There’s a very good reason we don’t test asymptomatic people in low incidence populations. Basically all positives are false positives when you do that, no matter how accurate the test is.
When you’re testing healthy randos for everything the odds of a positive being false have so many 9s it would make an SRE weep.
Unless this is accurate to a degree previously unheard of in medical science it’s a boondoggle, and I can’t help but notice there’s no mention of accuracy.
Unfortunately that’s just basic statistics.
To your point though I think there is a difference between collecting and evaluating additional data sources and using them as diagnostic tools.
I suppose I fundamentally disagree with the implication of your post that there is no value in gathering further data for these reasons, it would seem to suggest we’re already diagnostically optimal and could not do better with additional signal.
You [hopefully] won't have to become a rare missed diagnosis because you didn't fit the demographic for this or that screening test.
Cost of genomic analysis is exponentially decreasing, and so much progress is happening so quickly.
Consider for example how in cardiology we advanced from ASCVD's 10-yr prognosis, to the PREVENT 30-yr prognosis. And still most providers are using the ASCVD score for their patients.
I don't know about traditional blood testing, but a permanent implant which checks HR, pressure, glucose, temperature & oxidation would be pretty useful, not necessarily to diagnose anything, but to provide data for doctor when patient has actual symptomps.
It's news to no one that tests are imperfect.
Do you have any concrete solution to that? Anything of value?
This non-invasive everything-scanner sounds more like science fiction.
During Covid it was useful for improving protocols.
Somebody should make a startup based around the idea of diagnosing diseases through eg. a drop of blood. Probably need a bunch of big name investors though
On the other hand, nothing here substantiates this promise. We've got a video render of what a hypothetical device could look like. It's probably more than nothing (they got exclusive license on these butterfly chips in 2025, and it's at least plausible that the best solution to the data bottleneck in an absurdly noisy system like this is real-time AI image processing)... But it's certainly less than something. It's a hype video that doesn't prove feasibility of anything, yet.
EDIT: This is all in reaction to the second video on the announcement post[0], which is much more informative than anything on the page currently linked.
1. It kind of makes sense that an AI imagery company would apply that to other novel applications of imagery and computing and try to do something cool with it.
2. Midjourney as a brand is all over the place and this feels -off, somehow. I think from a branding pov they should have just started a different company with a different name. Perhaps a single image-focused umbrella company named [Name] with Midjourney and this medtech company as separate subsidiaries.
3. AI imagery companies suddenly making medtech products and spas feels very “we don’t know what to do, so we’re going to throw spaghetti at the wall.” That doesn’t necessarily mean it’ll be bad, just that it’s not typically what you’d do if you’re working on something super successful already.
4. AFAIK they are entirely self-funded and so this really isn’t about VC scaling or anything like that. But that doesn’t mean they’re immune to the same cultural pressures.
With "you" being a VC backed startup aiming for the next $1T IPO. What could possibly go wrong?
My only criticism from the tech video would be that they spend some time lauding the nanometer deflection sensitivity, which might lead some to believe that's indicative of the image resolution. It's not, and it's somewhat of a distraction -- that's just giving us amplitude information, which is comparatively less important than correlated time/phase across the 100k sensors. They do later on state ~mm resolution, which is still great!
Doppler and motion blur may be an issue (e.g. heart beating), as one slice requires a full ring of sequential exposures. But still way faster than MRI, so probably fine.
On a lighter note, it could seriously change the meaning of get FUCT (Full body Ultrasound Computational Tomography)!
There is a part of me that thinks it would be cool to get cheap full body scans. I like being able to see inside of myself. I can think of a lot of situations where the low-fidelity images coming out of this (they're not good compared to real medical imaging, if you've ever looking at MRI/CT up close) could be useful for coarse analysis of certain conditions that come and go or need to be monitored over long periods of time.
What I don't like is the idea of getting people to do full body scans every month just to be safe. This might sound like a good idea if you haven't looked at the literature on preventative full body imaging. Looking for bad things inside the body sounds like a great idea on the surface.
The problem is that imaging, especially when it's as rough as these ultrasounds, and possibly worse when augmented by AI guessing at what it's seeing, can lead to a lot of unnecessary procedures. The net effect can even become more harmful than the number of real problems it catches. There's a long history of research on this as many companies have tried to commercialize full-body scanning in the past. It frequently leads to situations where there's an unknown or ambiguous spot on the imaging that the person reading the scan can't rule out, which turns into a lot of anxiety and eventually more imaging, biopsies, or unnecessary surgeries. It's easy to think "better safe than sorry" until you realize how often these benign but ambiguous findings show up on full body imaging.
So my initial thoughts on this are that it would be good to make cheap ultrasonic imaging accessible as an as-needed service to use for specific conditions. I do not think it's a good idea to go down the road of trying to scan the entire population once a month and then run it through AI to see if anything pops up. The number of false positives would be overwhelming and lead to a lot of unnecessary procedures to calm the resulting anxieties.
Regardless, as a doctor and full stack engineer, I'm looking forward to learning more about their methodologies, their approaches, but I don't think this is going to be displacing MRIs or remotely close, based off the cursory initial glance. If their vision is to be able to provide end users with more actionable data with some kind of "low fidelity" medical imaging data that is somewhere above zero and or standard imaging and high fidelity modalities like CT/MRI, then this could be somewhat interesting.
Not a radiologist and not medical advice. Just my two cents.
I could imagine this getting cheap enough that your local gym has one and you get checked once every 3 months.
Curing cancer is one of the only things I’d take a pay cut to do.
Send an email to this head-and-neck oncologist's lab. I saw a talk he gave at a Chicago-area national lab on open-source models for identifying malignancies in scanned pathology slides, and was smitten.
Is this some AI hallucination post?
Yeah, that's not just 'cart before the horse', it's more like cart before the wheel. They make a bunch of extraordinary claims yet offer zero evidence, info or even a plausible hypothesis on how those claims might be possible at the scale, timeframe (2027) and unit economics implied. Thank goodness they really thought through the accent lighting for a calming user experience though. Otherwise, I might have been concerned they're not serious. </s>
https://cdn.midjourney.com/static/medical/media/first_mri_vs...
It’s already used in breast imaging (SoftVue) and hasn’t replace mammography. A body part ideally suited for ultrasound.
More compute many minimize some of the fundamental limits of sound waves (bone and gas) but I would be shocked if they have useful images of 90% of the body parts we image with CT or MRI and even beyond that I question how much it’s more useful than B-mode anyway.
Quite slow which means most things abdomen and chest will be motion degraded.
This may be useful in superficial areas but then why do whole body anyway. Might be some new niches and interesting research but hardly revolutionary in my opinion.
What there isn't is good evidence that these full body scans actually improve outcomes.
EDIT: Actually looks like their announcement has another page linked for more details containing this video.
The approach sounds like something which appears in a few research articles from the 2010s (ultrasound computed tomography), although submersion to make the ultrasound transmission more efficient seems novel.
It's possible the "spa" approach is used because it's hard to achieve the level of cleanliness required in a typical health facility using a shared bath.
Why don’t they approach this as a regular medical product?
With this spa angle I’m worried about hidden motives; perhaps data collection is a major goal. Or maybe this tech is not reliable enough.
Of course a lot of it is about the energy and overall exposure, and the harms of this, if any, are more likely elsewhere, but it's completely reasonable to question extraordinary promises made by people who up to this point have shown no expertise in the field.
I swear, it's like some people have already forgotten about Theranos.
Stuff like this needs to go through approvals for obvious reasons before they can advertise them for having medical purposes.
That's a tautology. We already have quite robust methods for detecting developed anomalies, treating every anomaly below standard human-to-human variation effectively raises the noise floor to already developed anomalies, defeating the purpose of population wide routine scans.
> The downside, and the reason why most doctors do not recommend full body scans, is that every human body is a bit weird and there will almost always be something "wrong" that will be visible in a full body scan. This can lead to unnecessary testing, anxiety, and even unnecessary procedures. Many of these oddities flagged by the scan would never have caused any actual issues had the patient never been aware.
The fundamental problem is that you generally can't diagnose simply from shapes. Scans show shapes, shapes cause concern, concern leads to invasive procedures, results are negative.
Also, overdependency on "spas" for health information could lead to an atrophy of other sorts of medical information gathering and diagnosis. e.g., there's no mention in the dreamy description of this spa experience of getting a blood draw or a patellar reflex test.
Is it? Linear No Threshold has largely been rejected at this point. https://jnm.snmjournals.org/content/early/2024/06/21/jnumed....
What's the relation between sensor density and resolution? If their array could give femtometer resolution, how much could you drop the density when you only needed to detect forearm muscle movements through the skim.
The way Ctl-labs was trying achieve the same results always seemed like it had fundamental physical limitations due to the nature of electromyography (to this software engineer...)
The safety of the device itself is a concern, but so is the trustworthiness of the output. Midjourney already has some very questionable history with medical imagery (like this totally legit image of rat testicles published in "Frontiers in Cell and Developmental Biology" https://upload.wikimedia.org/wikipedia/commons/c/cc/AI_gener...)
(I researched more and found in the video a value) The waves are 50 nanometres, and this is basically the equivalent of having a full body ultrasound. We've been doing baby ultrasounds for decades with no ill effects, so I can't imagine this being different
Could you expand on the term "working"? Do you mean like "working to slowly lower a person into water while videos of animated Figma UIs play back on a monitor?" Or do you mean some crazy kind of "working", like "the ring of devices we see are scanning the organs of the woman we see and the images appearing on the monitor are those just-captured organ scans?"
For sure, we have to be realistic about what processes will systematically have error, and if we can't stop a doctor from doing bad things with a piece of data we should shield them from it, but the tools to make scalable, calibrated risk estimates based on large data dumps is getting better every year.
There are physical limits to detection and technical parameters that make some situations indeterminate even for the best of the 'gud'. It is frustrating that, hearing an argument from many different individuals over a long time, you assume that each speaker is missing the critical insight that you possess.
> but the tools to make scalable, calibrated risk estimates based on large data dumps is getting better every year.
So your suggestion for indeterminate scans is more scans? There is no 'large data dump' personalized to you except for your own imaging.
> if we can't stop a doctor from doing bad things with a piece of data we should shield them from it
The doctor isn't the problem, it's the people who would be seeking out monthly imaging without symptoms
If Midjourney says "maybe you have cancer" but your doctor doesn't take it seriously, you might sue if you do end up with cancer. You might even win, regardless of whether "wait and see" was the right approach.
Meanwhile, if your doctor gives you an unnecessary CT scan that rules out cancer, hospital both earns $$$ and the doctor doesn't face legal consequences. Your increased chance of cancer risk from the radiation isn't something you can realistically sue over.
Exactly this. I mean, even if the scan is really indeterminate, at a minimum you can simply wait, then scan again. If it's truly something serious, it will become determinate at some point. Doing this is still better than nothing and carries no risks of unnecessary procedures.
If the scans are cheap and fast enough, the solution is to not do anything until you’ve observed the mass in question grow over time, not just be there.
So as opposed to bilking the ultra-wealthy to invest in a bunk idea, at worst this seems to be enticing them to pay for an at-worst expensive and possibly useless service. On that scale, it's downright ethical.
This project seems doable (just with a ton of data). Not sure about MRI level resolution, but CT is definitely not MRI level resolution but still extremely useful.
Their butterfly chips might be cool, but it's not like the article says anything about that. There's only one other comment in the whole thread that even mentions it.
[0] https://www.theinformation.com/briefings/midjourney-revenue-...
This community can be much better than that.
Remember, commercialization isn't the goal. They don't need to make a profit, as a company, they just need to get people to invest in their company and not get charged with fraud for something along the way.
It's like if LeBron announced he was switching to bowling and was going to revolutionize the sport, then rolled a gutter ball.
Never underestimate the audacity of a software engineer with a new toy
> It's like if LeBron announced he was switching to bowling and was going to revolutionize the sport, then rolled a gutter ball.
Well, if you replace LeBron with Jordan, and Bowling with Baseball ..
As a layperson, I'm mostly familiar with the concept of "get scanned, and a professional evaluates it"... are there scenarios where the approach of "imaging every few weeks, to make decisions based on trends" is currently done?
(From reading other comment threads here, I suspect the general answer is: other body-scanning startups have proposed the same thing, and it hasn't made sense)
As an aside, I could probably benefit from allergy shots, but the idea of having a regularly scheduled errand to do during the workweek is pretty unappealing, so I never seriously consider it.
Without those kinds of details, radiologists just expose themselves to: oh so you're telling me this doesn't work as well as the machines you paid ~millions of dollars for and are currently charging your clients a lot to use? Mmm I wonder why.
It's not clear that we have the health infrastructure in place to know what to do with frequent, low resolution, whole body scans of the human body. How often do anomalies show up and then go away? How often are anomalies purely a scanning/data processing artifact? Who reads the scans and makes recommendations about follow-ups, if any? I think this is the kind of thing that sounds exciting and with low direct risk, but with all kinds of questions that are not only unanswered, but apparently unconsidered.
The resolution of typical DICOM images is much less than what they're saying they are actually capturing, so the reconstructed images they're showing are just terrible for no good reason.
But I suspect there is a bigger fundamental physics issue with this entire thing... I'm not convinced they can penetrate fully inside and all the way around a human with only non-ionizing energy, especially from that far away.
If it has a 1% false positive rate but the incidence is 1%, the vast majority of the positives are false. Then you have to deal with the consequences, including invasive procedures for further diagnosis.
If you’re searching for tens or hundreds of low incidence conditions in the general population at a time it’s absolutely worthless because basically every positive is a false positive. At that point save the scan fee, spin a wheel of body parts and go get a biopsy of that.
This is why doctors are confused why companies are offering periodic full body scans in normal people. They only test people who are high risk or symptomatic to confirm a suspected diagnosis. That extra signal is what makes the test useful.
Go down to the medical diagnosis section for a worked example.
https://en.wikipedia.org/wiki/Bayes'_theorem
Regarding cancers every human has all sorts of weird lumps that are generally meaningless.
In order for this to not be a boondoggle it would have to be spectacularly accurate to a degree previously unheard of. Just from a statistics perspective.
This is exactly my thinking. There are decades of longitudinal studies behind the recommendations physicians make based on given levels of e.g. cholesterol in a standard blood test. And critically, those depend on standard protocols around administering and testing samples.
This would be brand new and would not have any of that infrastructure. Which all tech starts at, good. But I would expect Midjourney to need to dig in for a few decades to get and analyze clinical results and outcomes.
For body scans, I think about how few people would know if they have e.g. three kidneys (or other distortion), and how that impacts/doesn't impact their health.
Most people do not undergo autopsy after death, so it's possible there are correlates between good/bad health outcomes that frequent scanning would eventually reveal. But it would take significant time for this to be apparent.
> We’re starting by just giving you detailed body composition maps — and we’ll be submitting regular test results to the FDA for increased capabilities.
As far as I understand ultrasound there's no reason you couldn't do this, it's just infeasible to do a full body scan with a hand probe and you get covered in goop.
If the false positive rate is demonstrably low, I can't see the risk. People who think they need a doctor will go to a doctor with or without a fancy scan. People who want to play armchair physician will play armchair physician with or without a fancy scan.
The false positive rate is the entire risk.
When you go to the doctor for a physical they don't run all of the blood tests they can. They only run them for specific symptoms and for specific preventative measures where we've calculated that the benefits outweigh the risks of a false positive.
Some tests have been removed from routine exams, or at least discouraged, because they were producing more false positives and harm than what they were saving.
Full body scans are deep on the end of the spectrum of tests with high false positive rate when ordered without supporting symptoms. That's the risk.
> People who think they need a doctor will go to a doctor with or without a fancy scan. People who want to play armchair physician will play armchair physician with or without a fancy scan.
Not really how it works in real life. When you get a full body scan, especially with ultrasound, there are a lot of benign things that can show up that vaguely look like non-benign things. Even if the interpretation is "probably nothing", many people start worrying and think they need to get more tests just to be safe. Even people who don't see themselves as "armchair physician" will start thinking that they should at least rule out the worst case because they wouldn't want to die of cancer having known that something might have been there.
A new chargeable procedure is for for the hospital but maybe not for patients imo.
> So your suggestion for indeterminate scans is more scans?
The solution to imperfect evidence is consistent and calibrated risk estimation of both disease and intervention.
The trick seems like it would be to strongly incentivize waiting and watching any symptomless anomalies if further investigation is invasive. If you're getting 60 second scans every month then something growing will be catchable and something static or that disappears can be ignored until the next scan.
If you mean run different tests, where you collect different kinds of data from the same individual, sure but that's not something you can "just do" in the general case.
More seriously, I assumed that CT Ultrasound image is from Butterfly's actual FDA-approved handheld medical device, not the Midjourney 360 submerged ring - as there's no evidence that is working. Since the Midjourney site has no helpful information, I just asked a friendly AI to do a comparison of what's actually proven to work in the Butterfly chip which Midjourney licensed and this 360 degree, full body, submerged concept - and essentially what's not been proven to work are those three differences: 360 degree ring of 40 butterfly chips, full body at once (requiring solving distance and speed challenges as well as a massive signal processing problem to extract and denoise signal), and doing it submerged.
More often it leads to people thinking they have issues when they don't.
The same thing happens with blood tests: You can order all the blood tests you want if you're willing to pay for them. If you order enough, you will get some that show up as abnormal. You can start spending tens of thousands of dollars ruling things out and never catch any real issues.
Every system that exists as a black box is more understandable with more sensing, not less. Our bodies are not special.
It's also ridiculous that the proposition goes like:
1. Doctor knows some tests will flag tumors or variations that look weird and that we shouldn't then go investigate all of them
2. Doctor shuts off their brain and will then investigate all of them by doing invasive procedures
Just knowing how many such variations there are and if they grow or not is useful information. But the doctors pretend like they are super smart before the test and super dumb right after.
And the safety of the data as well. Am I supposed to entrust full body scans to a startup?
The chance a positive is real is so low you may as well just point to a body part and get it biopsied.
A positive from this kind of test is statistically meaningless.
If you let it give out tons of false positives, then patients are trained to ignore it when it cries wolf.
If you dial it back so that it gives out fewer positives, then now it starts giving out false negatives and not helping sick people.
https://www.rieti.go.jp/en/columns/a01_0455.html This japanese article found "No clear-cut evidence exists to determine whether undergoing health checks leads to greater longevity and/or lower medical expenditures."
https://pubmed.ncbi.nlm.nih.gov/31642821/
And blood pressure is especially pernicious, basically every doctors office measures it wrong so the results aren’t particularly useful. Many use the wrong size cuff for example, or don’t give people time to relax before a reading. A ton of people have white coat hypertension, high BP only because they’re in a doctors office.
https://pmc.ncbi.nlm.nih.gov/articles/PMC1120072/
I saw a paper that showed only 36% of cardiologists did it right.
Let doers do.
Biopsies are expensive, waste time, hospital resources and carry risks of infection and scarring that do not net out positively for people who aren’t in your risk group.
Getting a totally random positive doesn’t put you into a higher incidence category so whatever follow up test you take will be just as inaccurate as the first one.
The reason to avoid them is the tests would be a waste of time, statistically, and expose you to a bad risk-reward profile.
If you knew apriori 99% of the positive tests are false positive why are you taking the test?
It’s literally just math. Sometimes the right thing for you on average is to do nothing, which feels bad, but it’s still the right thing to do.
A good primer: https://pmc.ncbi.nlm.nih.gov/articles/PMC2477686/
