I own the 1st gen of the Butterfly - in my opinion, it wasn't great image-wise compared with the contemporary conventional crystal based probes (thinking of cart-based machines with less flexible, more expensive probes etc so perhaps an unfair comparison). Would be cool if the newest ones mentioned in the article are becoming comparable with the crystal based probes - I can't comment. But I can say image quality is absolutely key. There are lots of cool AI based applications coming out all the time (I know much more about echocardiography AI than the foetal ultrasound AI mentioned in the article, but this is a similar paper where some ultrasound novices had AI guidance and were able to obtain useful echo images https://www.ahajournals.org/doi/10.1161/CIRCIMAGING.123.0155...). I get to use various machine vision based tools on echo images at work to automate various measurements - but at the moment, I find they fail badly if the imaging is anything but great quality, whereas humans can interpret them. Maybe future training sets will include more "technically difficult studies" (code for poor imaging) and AI tools will do better than they do now? Or there will be more augmentation of data sets with realistically degraded versions of images to add robustness? AI that worked on suboptimal images would be awesome, particularly in my setting (ICU).
Medical imaging is not about getting the expected result as is typical in most bodies, rather, it is about getting the actual result as is found in this body.
The applications I've seen are around analysis/alerting. Submit your raw data to a service and it screens for hundreds of rare conditions your doctor's never heard of.
Similar to blood work maybe nothing comes back conclusive but it might be interesting enough to trigger follow on investigation.
But the general theme is around augmenting the human decision maker rather than "preprocessing" the data in a way that might obscure or hallucinate important details.
False positives on tests are a well understood source of illness and iatrogenic harm, as well as being a well known money printer for those administering the tests. Trying to screen for a whack of rare diseases with AI still sounds dubious to me unless you have good reason to suspect one beyond high levels of medical anxiety.
I’m more interested in AIs ability to do things like read x-rays and identify illnesses with higher accuracy than a tired doctor on a 12 hour shift.
Sure, there are many applications and more coming. In your example of an actual result as is found in this body, the image segmentation [0] and recognition task is applicable.
One of the bullet points in the above is "Intra-surgery navigation." Let's say there is a known position of a tumor within an organ based on previous imaging. The human body is squishy and interior organs may shift in position or shape when a procedure is performed. Deformable image registration [1] guides the surgeon (human or robot) to the tumor location with greater precision than what was possible before.
Worth mentioning another nice initiative https://www.echopen.com/ it’s from Paris and it has been incubated by the hospitals of Paris themselves, first as an open-source project.
Disclaimer: I was among the early contributors to the open-source project.
Last year I had arm surgery and in the recovery room I was in pain and agreed to a nerve block. The anesthesiologist put an iPad in my lap. Then with the ultrasound head in one hand and a syringe in the other he looked at the iPad until he found the nerves he wanted and zap!
I was in a lot of pain and was a bit groggy but my one clear memory is being fascinated by this and watching the tablet closely.
(The nerve block worked, but I don’t know that I would ever get one again. It definitely freaked me out).
Laptop-sized ultrasound machines have been a thing for at least a decade at this point. It feels insincere to frame ultrasound machines as these big things lugged around on carts still. Hospitals still like those, but there's plenty of pre-hospital ultrasound machines that are compact. Not smartphone compact, but it feels weird for the article to completely omit them.
The other thing to consider is that these machines cost money and have a write-off lifetime, I wouldn't be surprised if they're designed and purchased for a lifetime of 10, 15 years, with only some consumables like new probes and cables every once in a while.
The article is a bit of a puff piece for Butterfly and Exo. MEMS address a problem, but not necessarily the hard one for portable ultrasound. There are plenty of handheld plain old PZT transducer based devices out there.
Me too! That's the only reason I know laptop-sized portable ultrasound machines exist. I had an internship there one summer where I wrote automation software for the M-Turbo.
Having just had an ultrasound of my prostate, where the probe went up my ass, I welcome smaller ultrasounds!
I was able to watch the procedure on the screen. The ultrasound image was overlayed on a prior MRI image, and it was used to guide the biopsy where 12 core samples were taken of the prostate with a needle. It sounded like a staple gun each time a sample was taken!
Tiny Ultrasounds will be very useful for all sorts of biopsy procedures like this...
Very interesting overview of the technological / materials breakthroughs to enable miniaturization.
From a physician (specifically radiologist) perspective, I'm a big fan of handheld/point-of-care ultrasound and am excited about their potential at democratizing a very useful and low-cost/low-risk imaging technology. (I also own a Butterfly.)
That being said, the "cart-based" ultrasounds will likely always have a place in a hospital; the size of the ultrasound probes is not why the cart is big/expensive/useful. The cart is a big floating screen and also an image/record management system. You (or at least, a trained ultrasound technician) can manipulate ultrasound parameters and annotate images (critical for ultrasound interpretation) way way faster on the cart's set of wonky keys/dials/trackball than on a smartphone.
Also, with the rise of handheld/point-of-care ultrasound, we've noted with amusement in the radiology department the frequency with which patients get referred for additional imaging because the ED or primary doctor saw some pathology on their handheld ultrasound... and when we take a look it is just not there. I think this is probably more of a training issue, as ultrasound is truly quite challenging both to perform AND to interpret (and a major part of using an ultrasound probe is essentially real-time interpretation), which is even more challenging given the lower image quality and field-of-view of smaller probes.
In radio we are seeing phased arrays and software defined systems becoming more popular.
Why can't we expand the resolution and increase the bandwidth with ultrasound as well using the same ideas.
Wouldn't we be able to get a more clear tomographic pictures if we used many receivers places around the body and layered different frequencies? I'm especially wondering why not use lower frequencies for deeper penetration, using phased arrays to offset the loss in resolution.
A key advantage of US over other imaging modalities is portability. A key disadvantage is resolution with depth. Tomographic ultrasound systems exist, but if one is going to go with a bulky machine you might as well go with CT or MRI.
For what procedures are the current resolution limitations inadequate? What would be the cost difference towards a machine that provides adequate resolution?
Couldn't find anything in the article about whether these still require gel (or if the gel is an absolute requirement for the conventional ones). Anyone know?
Now here's an application for "AI": have a ultrasound device the size of a medical tricorder from Star Trek in every household with children. If child gets one of the dozens of inevitable infections, use the device to check for the standard complications (otitis media, pneumonia ...) and allow a physician to replace 95% or more of the visits with a remote diagnostic.
Agree in principle with remote diagnostics in this way. But, the number of acute illnesses that can be diagnosed with US and can/should be treated at home is fairly small. Pneumonia cannot be detected by US unless it is up against the chest wall (or perhaps if it causes excess fluid in the pleural space). An otitis media diagnosis would not be possible with US. Appendicitis can be seen with US but it is very operator-dependent and asking parents to fish around for the appendix would be a big ask (even with some rather advanced AI assistance). You can certainly find enlarged lymph nodes but fingers are often just as good.
My understanding is that's not how ultrasound works?
CT and MRI give pretty clean images that even a layman can understand. I have no problem identifying things on them.
Ultrasound is looking for a vague voxel of density that is or isn't where it's supposed to be. The implies that you need to know what the density looks like when it's normal vs abnormal. It's more like "Your liver is supposed to be there but there is a blob of water instead. Let's get you a CT scan." If it's abnormal, then you go confirm with a high-resolution scan system.
This has a LOT of room for error. Ultrasound is how they identify gallstones, but, in my case, the ultrasound said "Yeah, you've got some gallstones, but no big deal." When they removed them after my gallbladder exploded, they were almost an inch and a half across. The problem is that gallstones are mostly just globs of cholesterol that isn't much diferent in density from the surrounding fluid.
Ultrasound is okay, but cheaper, smaller, faster CT/MRI systems that don't need contrast dye would enable far better diagnosis as it would put imaging much closer to office physicians.
Of we can get AI to produce good quality images from ultrasound sensors and then to interpret the images, we can reduce the level of training required to use such devices.
Imagine high school and junior high health classes teaching the use of these devices, getting the students certifications that require maybe 8 hours of coursework plus a 30-minute exam, then sending most of the clas home with a 21st century tricorder.
There's already a "pandemic" of incorrect and/or hypochondriac "at home" diagnosis going on, along with quack pseudoscience to cure things, ranging from essential oils to bleach enemas; we don't need more of that.
This is also an answer to a perceived problem, mixed with the XY problem; the problem isn't that there's not enough at-home diagnostic tools (a thermometer is enough for most households), the problem is access to health care, which has been messed with directly (cost of health care) and indirectly (people's physical health due to overconsumption, mental health due to gestures everywhere vaguely) by capitalism.
> There's already a "pandemic" of incorrect and/or hypochondriac "at home" diagnosis going on, along with quack pseudoscience
There is also a revolution in hone medical technology.
It takes a long time for us lay people with cool medical gadgets we got off eBay to learn what the results mean, but we will get there.
If fed.gov wants to help, they could test these devices, then offer training and certification so those who take advantage of the certification classes will know how to get reliable results and (more importantly) what the machines can not do.
I remember a hot discussion here and only a few years ago where an "expert of the field" said small cheap portable ultrasounds were impossible, despite one or two commercial products available at the time.
This is wild.
Also if this becomes cheap enough for regular folks to buy I can see a lot of self diagnosis coming in. Then again, maybe that'll be good
Good for what? How often do people need an ultrasound, and how much would it offload the existing medical whatnots? It would actually increase the load on the system, because people would be like "My ultrasound detected a thing, I need someone to look at it".
I’m not convinced self diagnosis will be a positive thing.
What these small, portable machines will support is a lot of pre-hospital medics, working in remote or austere environments, to carry these devices and use them effectively.
There are already courses available to teach exactly this:
You're not allowed to buy these without an NPI number. And while ultrasound is relatively easy, it's still not really usable without some training. I think there have been some studies of training users to do ultrasound at home for a limited set of views, but it's not really a "pick it up and look around" sort of thing.
I doubt it. How can you image an organ at home when you're not even 100% sure exactly where that organ is? Or what it looks like on an ultrasound? And that's not even getting into knowing whether you're looking at them from the right angle for the images to be diagnostically useful.
Yeah - my experience of watching baby ultrasounds was that I had no idea what I was looking at until the technician told me. Maybe there's an occasional clear image of a head or nose or something, but the rest of the time it's just wobbling grey shapes. I can't imagine trying to do that myself - let alone looking at my own organs which are mostly just amorphous blobs, and trying to diagnose something.
I suspect the main cost in running a diagnostic scan is labour rather than capital - and these low cost ultrasound devices will only have an impact in the developed world if they go hand in hand with AI.
Obviously they'll make a huge difference in developing countries where labour is relatively cheap compared to equipment - but even then it'll still require trained professionals to operate.
A bit the same with infrared. About 10, 15 years ago, several startups came out with cheaper ir-cameras. Flir then came out with the first gen Lepton chip, lowered the price for a while, so they where cheeper than the competition, i think none of the startups survived. Now you pay again a much to high price for a thermal camera ..
While the article states that it did not include patient data, people living in the UK a few years before that had to opt out of their data being given to Amazon. Imagine that.
It's different from the MEMS-based devices this article talks about, which have the novelty of letting you do everything with a single probe. Though of course that comes with trade-offs.
I'm sure there's a time and a place for these things, but it cannot be said that something as innocuous sounding as 'ultrasound' is 100% safe. It has an effect, it can be used to break up kidney stones for example. I personally wouldn't use it for scanning pregnant women, despite this very common usage.
Turning it into a small consumer level device sounds like it could lead to people causing themselves harm.
It is dose dependent. Use of low powered US for a short period is safe. If you increase power, you can cause damage - either to to break kidney stones as you say, or, when focussed properly, burn bits of the brain to treat tremors.
It depends on one's definition of "safe". Yes, smaller doses will have less of an impact.
I'm simply stating that ultrasound does impact the body. Despite the innocuous sounding name referencing 'sound', it is less like music and more like a microwave, capable of breaking kidney stones, burning tumors.
Re the most common usage, I don't think lots of scans in pregnancy can be beneficial. Perhaps one or 2 scans are worth the risk in special circumstances. Present day scanning machines are way more powerful than historical ones; the dose is not getting safer.
Capable of, but the ultrasound used for that is focused, higher intensity and in bursts.
I mean if you want to fearmonger, there's plenty more examples of things that can cause damage if used in a different way, like water, air, electricity, light, etc.
You seem to want to claim that any use of ultrasound is dangerous, but you fail to provide evidence, you just bring in extreme cases and try to claim equivalence. I'm sure there's a fancy sounding fallacy name for that.
What I am saying that it is agreed that ultrasound does have an impact.
The failure in the provision of evidence, is with those that claim it is safe. Its not for me to prove it is harmful.
Ask yourself, how can a treatment that breaks kidney stones or burn tumours also be 'safe'? It may be that the impact is negligible, but that is not safe. In the womb especially, a slight impact may have a large effect on a growing foetus.
Ask yourself too, what level of radiation is safe? How much microwaving? Is any?
So am I fearmongering, or is this a valid concern? I think it is clearly a valid concern. I do not simply accept medical consensus over the reasoning of my own mind.
I can see use cases where the information that is provided by ultrasound is useful and outweighs the risk. I think that casual, repeated usage of ultrasound for pregnant women, for the 'gender reveal'/colour scans or whatever, with powerful machines, is ill advised though. I don't recommend anyone undertake any unnecessary medical procedures.
I also think that an adult that has awareness of the scope of a potential impact, can do what they want - its their choice to do as they want. I object to inflicting avoidable harm (albeit the harm might not be obvious) to individuals that cannot consent - eg infants, foetuses.
FWIW, I do have kids, and - to my regret - I did scan them before they were born. I wouldn't do it now, knowing what I do.
>FWIW, I do have kids, and - to my regret - I did scan them before they were born. I wouldn't do it now, knowing what I do.
Why are you acting like you have scientific consensus behind what you're saying? There is literally zero proof of anything you're claiming.
>So am I fearmongering, or is this a valid concern?
You're fearmongering. That's it. Nothing else. It's the same kind of attitude that has led to such a self-important, 'I know better than doctors' attitude about vaccines.
> The first mention that ultrasound could be used to produce images of the foetal head was probably in a lecture given by Ian Donald in 1959. Since that time, the use of ultrasound in obstetrics has grown rapidly, and has a generally accepted excellent safety record. However, it is impossible to prove zero risk, and the absence of evidence of harm should not be taken as evidence of absence of harm. The epidemiological evidence that exists is reassuring as to the safety of routine ultrasound scanning, but of necessity it only includes subjects who were imaged with devices that were state of the art at the time (mostly early 1980s). No pulsed Doppler or colour flow examinations are included, and the output from modern ultrasound scanners is considerably higher today than it was at that time. It is, therefore, essential to remain vigilant, and to assess new technologies and applications from a safety aspect as they arise.
So TL;DR: Ultrasound is safe, the benefits outweigh the risks by a large margin, and we should always insist that Ultrasound, and any modern medical technology is held to rigorous scientific testing standards.
Listening to music on a little portable speaker is perfectly safe for your ears. Standing in front of the stack at a 500,000 person concert is quite harmful. These devices are on the “portable speaker” end of that spectrum.
Smellovision; that said, there's people and/or dogs that can smell specific diseases, if they can figure out how that works they can incorporate it into an array of passive sensors to alert people of a possible problem.
