Hacker News new | past | comments | ask | show | jobs | submit login
FDA clears ‘world’s first’ portable, low-cost MRI (healthimaging.com)
447 points by apsec112 on Feb 16, 2020 | hide | past | favorite | 130 comments



For a change, I can contribute something to HN without being out of my depth:)

Background: I'm a doctor with 3 years experience in acute medicine. AMA

Clearing some MRI misconceptions: 1. It's indicated ASAP in specific emergencies and will change how we manage them. For acute ischaemic stroke, it detects patients within the recomended time window for thrombolysis. Simply, it shows the cellular swelling in the brain, gives an estimate of the onset and we decide the risk/benefits of dissolving the clot. Second use, for cauda equina syndrome. Again we're looking for acute CNS tissue damage, this time from spinal chord compression, and the change in management is emergency neurosurgery (good luck getting them out of bed without an MRI) For herpes encephalitis, MRI is debatable because you can start empiric treatment. But I've seen it done.

Another misconception for the first use scenario in stroke. The MRI sequences we're interested in only take 1-3min and CT doesn't help us. But in my experience the whole scan takes 2h to organize, so point of care MRI is a game changer. Every minute counts when you're saving brain tissue. And Lower definition with artifacts are acceptable because you're looking for gross changes in a critically unwell patient.

Bonus: A report from when I managed my first suspected stroke as a junior doctor.

"49yo lady day 1 post op for total knee replacement (elective admission). Commorbidities of hypertension, obesity and osteoarthritis. She was just started on apixaban and gabapentin as per protocol, when I was asked to see her for new onset upper limb bilateral tremor.

On examining her I found the symptoms had started 1h ago. She was presenting bilateral upper limb ataxia as well, reflexes were reduced on the left and there was numbness corresponding to C4 dermatome. At this point I was worried about acute stroke whithin the 4h window for thrombolysis. I discussed the case with the acute stroke consultant and examined the patient with him. We requested an MRI brain to assess for acute posterior circulation stroke, because he thought time of onset was unclear (as per WAKE-UP protocol). I acompanied the patient to neuroradiology services. The MRI sequences we were most interested was DWI and FLAIR which only took a few minutes. We quickly scanned through the images noting there were was no DWI-FlAir mismatched high intensity signal areas within the brain parenchyma. This allowed us to rule out acute ischaemic stroke.

6h later the symptoms subsided and I was satisfied they were due to an adverse reaction to gabapentin."


Being a neuroradiologist (at Karolinska in Stockholm, Sweden) I really love the idea of this and almost can’t wait to start doing comparative studies to find how sensitive it really is. As I see it there are a couple of areas where it could be of tremendous use - first of course the (neuro) intensive care unit, where patients sometimes are to unstable to transport to the MRI, and where every transport is a potential risk for the patient, and where it can be very useful to find out whether they have ischemic lesions in the brainstem (which can be difficult to rule in/out on CT) or extensive diffuse axonal injuries, etc.

Another area would be pediatric imaging, where it could be useful as a first imaging, ruling out larger lesions, possibly lessening the need for CT and decreasing radiation exposure, especially in the group approximately 1-4 years, who often need sedation to lie sufficiently still inside the MRI. Although at my institution the physicists have developed a ‘fast MRI’ (70 seconds), that gives reasonable resolution and contrast (T1,T2,T2*,DWI), and which we will try to implement in the group of patients who come to ‘try’ the machine (before deciding whether they need sedation or not).

Also, the low field strength, 0.064T vs 1.5T or 3T, would most probably allow us to image patients with implants which are unsafe at normal clinical field strengths of 1.5 or 3T (would have to be verified though).

For people interested in low field imaging the group at Athinoula A Martinos Center for Biomedical Imaging at Harvard has a homepage at https://www.nmr.mgh.harvard.edu/lab/lfi


Thank you (and the grandparent poster). We (or at least I) love seeing more HN participation from insightful professionals from outside the tech industry!


The combination of the cost and accessibility look like a game changer for me, a lay-ish person.

With such a device, do you need a Radiologist or other rare super specialist to actually read the images? Or can they be used by generalists?

Will this turn brain MRIs into a tool comparable to the X-Ray scanner my dentist uses or the Ultrasound used by the midwives a dozen times during pregnancy?

If so it looks like a game changer - something which will become a fixture in A&E departments everywhere.


You are of course right about the cost and accessibility, low field not necessitating specific rooms or ‘no go’ zones, and presumably easy operation (controlled via iPad app) being very interesting with regards to increased availability of MRI scans(in Sweden the rate limiting step at the moment is usually the availability of trained MRI technicians). In my initial comment I mostly thought about some clinical situations in my current practice where it would be useful. Sorry if the following becomes a bit rambling.

Reasonably there are a plethora of indications and given that these machines will be installed at the point of care (A&E, ICU, etc) and supposedly are safe to use on basically all patients, one can imagine that it will be used very frequently. Though in this case the scan-time (30 min + change) could become a limiting factor (but, given it’s price, I guess you could just buy more machines). The image quality is of course not the same as with a modern clinical system but it is most certainly good enough for a preliminary test, and may very well decrease the number of patients sent for a ‘normal’ diagnostic scan (though the opposite may also be true, if you see something unexpected in the images, or if there are artefacts which are difficult to interpret). As with the ‘fast MRI’ I referenced to earlier, my current idea would be that it can be used as an initial screening, but, as always, if the test is normal, but the clinical suspicion persists one would have to continue with other tests.

Regarding whether the image should be read by a specialist or not, I am of course biased towards the specialist :) As I see it the main advantage of having the clinician interpreting the image is that they have ‘direct access’ to the patient, and can ask them specific questions based on the findings of the test. Though, time-utility wise it’s faster for a specialist to read a normal test. During med school I did a rotation at a primary care facility in the north of Sweden, in a municipality with 6000 inhabitants, located almost 2 hrs away from the closest hospital, the GP:s there read chest x-rays themselves at that time. Now the images are transferred digitally to the main hospital of the region and read by a radiologist. One can of course wonder whether it’s cost effective to train a large number of radiologists to read all the new studies which will be produced. My hope here is that computer aided diagnostic tools will increase the productivity of radiology specialists.

My experience from working together with neurosurgeons and neurologists who are very good at reading images in general, is that they anyway prefer to consult us for the interpretation, (they would rather operate than become specialist radiologists). In my opinion the professionals at risk of being side-stepped are the general radiologists, and I think this is already happening as telemedicine makes it possible for smaller hospitals to pay for specialist readings on a per study basis.

edit: regarding time utility, one would of course have to include the time it takes for the referring physician to write a request, and for the radiologist to write a report. And another advantage of having the clinicians read the image is that they have more information about the patient than they put in the request. So it’s also very possible that radiologists will not be consulted (depending on medicolegal circumstances and reimbursement systems etc.)


My experience as well is that there's a relationship of trust between radiologist and surgeon that a trained algorithm cannot replace. Same reason why AI isn't a threat to sales people for high priced items.

I'm actually doing what we call a taster in neuroradiology and it's been fascinating. Strongly considering applying for residency next year!

I did my elective at the PET center in Turku, great team. They even let me play with the rat MRI for Alzheimer's. They had some cool projects under Marco Bucci at the time, I believe he just started at Karolinska.


Interesting, thanks for getting back :)

I worked for a couple of years in medical imaging (on a RIS) and our customers were very Radiologist focussed for everything except head scans (where the neurologists also read the scans).

I would hope to see AI-based assist technology hitting the market in the next 3-5 years for X-Ray/CT. Something in the form of the different rendering modes that you can use in a viewer. It's hard to tell though because the whole field is driven by hardware not software. Software seems to be an afterthought. The software seems more focussed on lock-in (integration, meh).

I had a vision for a future where the doctor in a hospital could push an image out to the "cloud" of teleworking radiologists, with the image being read in real time. It would be like load-balancing all of the work over the entire world. Technically it's tractable but the combination of privacy issues, the aforementioned hardware-driven market and the state of Hospital IT is what makes it impossible. Well that and the scaling problem (you need a critical mass of top-quality teleradiologists on your side to do it).


Yeow, 30 minutes?

I'd say the couple of MRIs I've had done took around 10-15min or so. I think I shifted a few mm slightly at one point (possible since you're not quite bolted down, just held firmly). For 30 minutes I'm quite sure I'd probably shift around quite a lot, either involuntarily, because I got uncomfortable, or because of mild claustrophobia. :/


> With such a device, do you need a Radiologist or other rare super specialist to actually read the images? Or can they be used by generalists?

Non-radiologists already see all the relevant imaging, they’re just not nearly good enough to read them. More machines won’t change that.


Will ER rooms need to be re-designed to accommodate bedside MRI? The ER at the hospital I work at already has special rooms for Gyn and Ophtho complaints so it wouldn't be that far fetched for there to be a room designed for Stroke code patients.


gabapentin gave me full on Parkinson’s symptoms. They would take several days to hit, and several days to resolve from a single pill. So it took a couple months to figure out what was going on.

Doctors had no clue what was going on.

Some of these meds are beyond scary.

Kids still wake up from nightmares of having family and teachers preparing them from what seemed a likely funeral.

See a different doctor and explain symptoms. “Oh yes, it some times does that.”

Getting correct diagnosis and meds is such a tricky topic.


It says the clearance includes head scanning, but could this also be used for things like orthopedics? Current MRIs are prohibitively expensive and bulky for all kinds of things outside head trauma that could benefit from tissue imaging.


Oh orthopods, they're a riot in the Uk. They won't have much use for a portable MRI. In trauma all they see is the bony bits where CT and XR are king, anything else is someone else's problem. A true ortho emergency is rare and patients can wait for surgery if neurovascular intact. They do have time guidelines to operate neck of femur fractures (still won't operate out of hours though). But an XR is often enough to proceed . A few caveats I can think of... For fulminant infection like necrotizing fasciitis: straight to OR, open and have a look at the tissues. Joint infections: stick a needle in and await results. Bone infections benefit from MRI but there's no change in management while stabilizing on antibiotics, plenty of time to arrange a scan.

Then we move on to elective surgery (planned). MRI with the best available resolution will increase diagnostic specificity and avoid unnecessary operations to labrum's, meniscus, etc. They don't need to be done in hospital and you can shop around.

At this point we're looking for problems to fit this solution, poor business model. Not surprisingly, limb only MRI machines haven't been successful.


Fair points, but my thinking was more for cases here in the US where insurance may or may not cover an MRI of your choice for something like torn rotator cuff, or where you'll end up paying your full out of pocket maximum for it. Cheaper options in such cases would be welcome.


> For herpes encephalitis, MRI is debatable because

I’m curious kind of institution you’re at? Everywhere I’ve been (NE) it’s fallen down academic lines: adverse impact of unneeded antiviral tx vs. mortality benefit in hsv encephalitis is incredibly one-sided, and time sensitive. Academic tertiary care institutions basically considered it borderline malpractice not to just initiate antivirals and d/c later. Non-academic secondary care institutions agreed in theory, but basically only started antivirals if everything else was off the table and/or imaging was characteristic (aka, either too damn late or later than it should be), but cases were rare enough it didn’t mean much in practice.


NE as in North East? I'm in the same region at a small academic center.

Saw 3 suspected cases recently, all started acyclovir within 1h in addition to the antibiotics, while awaiting lumbar puncture.


From what I saw of this at RSNA, they were only demonstrating head scans. The scanning volume is big enough for a head but not a full body. The magnet is low-power enough that you could walk within a few feet of it during operation without danger.

Cool device which could make MRI scans more accessible, and also paves the way for specialized scanners that only focus on a particular body part. I like medical tech that does a "good enough" job in exchange for being much cheaper than fully featured devices.


This is not technical nor attempting to be glib.

I've been to the RSNA, am involved in healthcare software in the USA, and seeing what companies like Phillips, GE, et al, spend on floor space at this conference, it seems difficult to imagine these companies yielding their market position for cheaper machines to reduce costs. MRI equipment is huge business (these companies don't bat an eye spending $1,000,000+ for a conference such as the RSNA because this is where they wine and dine and make every hospital exec feel "special" so when it comes time to buy x MRIs, the sale alone will pay back the conference cost 10 fold. These conferences are expensive. The sales pipe is long (and thus expensive).

There may be companies creating and selling these, but the heavyweights don't like losing a market opportunity in "their" hospitals. I'd be very reticent to think there will ever be lower costs in the ER for offering said device/service.

Please don't get me wrong, I'm excited for limited use machines that can provide the basic information needed at a fraction of the price. I'm very leery of US healthcare institutions making a rational, patient-friendly cost conscious decision.


These won't replace existing MRI machines though. These will be an additional purchase by hospitals or purchased for small, rural hospitals that would never buy an MRI machine in the first place.


Something that fits a head will also fit a hand, foot, arm, leg, every extremity except torso. The benefits are tremendous.


Yeah specially for people that cant sit still in the mri machine. My cousin is claustrophobic he just couldn't stay still enough they had to sedate him in the end.


Upright MRI's are sometimes used for that (claustrophobia). Some might be available nearby. They are not quite common but not that rare either. Depends on the scan as not everything can be done as far as I know due to lower magnetic fields.


If you're looking for this, the key word is often "Open" MRI.


That part is harder than it seems, good quality imaging of many of those locations is hard with a body coil. So the fact you can physically fit a leg in, say, doesn’t give you usable knee imaging. There also may be issues with the shape of the field if it is too tuned to cranial imaging, but the center should be pretty flat.

I’m not saying it couldn’t work with a bunch of extra RF engineering, but it’s not a given. For that matter some locations mist don’t have very good mr imaging regardless of setup.


Having a couple of these around might also free up the full sized devices for scans that require a full size device.


I was thinking that inexpensive devices might lead to MRIs outside of the 1st world.

quick search shows in 2016 USA has 37 mri/1mm people, mexico has 2 mri/1mm people


This would still be hugely helpful for brain imaging en route to a medical facility, similar to what’s being done with experimental mobile stroke units, so medical practitioners are ready to act upon a patient’s arrival.


Maybe. MRI is not the preferred brain imaging modality in emergency situations, X-ray computed tomography is.

Consider patients who are unconscious (or awake but unreliable) due to head trauma, stroke, etc. You can't do a proper MR safety screening interview to find out if they have ferrous metal or other conductive objects in their head. The magnetic field can potentially torque an old aneurysm clip or piece of shrapnel, and the radio frequency excitation field can cause heating and burns if there are antenna-like conductors present.

MRI is slow and the images are easy to ruin if the subject moves around too much during the long acquisition time, so bouncing around in the back of an ambulance isn't the ideal setting. A 3D volume with whole head coverage and suitably fine spatial resolution can take several minutes to acquire, versus just a few seconds for X-ray CT. The MRI signal to noise ratio is also proportional to field strength, so low field systems like this are at a disadvantage for image quality. On the plus side, low field does mean less risk of moving ferrous material inside the subject.


https://www.aan.com/PressRoom/Home/PressRelease/849

> According to the guideline, diffusion MRI should be considered more useful than a CT scan for diagnosing acute ischemic stroke within 12 hours of a person’s first stroke symptom. In one large study, among others, that was reviewed for the guideline, stroke was accurately detected 83 percent of the time by MRI versus 26 percent of the time by CT.

Main thing they need to know is stroke vs aneurism and accuracy is really important.


> You can't do a proper MR safety screening interview

Are we really not capable of detecting such dangers externally without patient input? It seems like detecting even small pieces of metal would be relatively trivial.


Presumably with an X-ray/CT scan.


Exactly! Maybe my initial comment should have emphasized the relative significance of reasons for preferring X-ray CT over MRI in an emergency:

1) Speed.

2) Depiction of bone (less relevant if it's not a trauma case with suspected broken bones).

?) Risk to patient in case of internal metal.

Some MRI suites are set up with airport style metal detectors to prevent people from accidentally bringing metal into the magnet room. I haven't heard of metal detectors being used to screen non-responsive patients, though.

It's a reasonable question, but I think that even if some instantaneous metal detector with a false negative rate of 0 could be made, emergency docs would rather scan for five seconds to get a CT than five minutes to get an MRI that has coarser resolution, poor bone depiction, and a chance that the whole image is ruined if the patient moves.


Well, obviously.

What I'm saying is, why isn't there a sensitive enough metal detector which outputs a binary value of whether or not a dangerous level of metal is present? You need to check for two things I imagine: ferrous objects which could be pulled by the magnet, and conductive objects which would heat up when absorbing the MRI RF pulses. Both seem possible to do with a bedside device.


Can’t you apply a minuscule power and watch out for overcurrent?


Some devices are safe to scan at lower intensities and for less time, but it's the radiologist who check on that, so not my expertise.


No, the main field is effectively permanent.


I mean, some sort of non-super-conducting MRI analogue device that the patient will have to go through.


You could find big things that way, but maybe not small things that were deep. I suspect it is hard to make something powerful enough to detect everything that might be an issue in say a 7T field, yet still safe.


It has applications in emergencies[1].Speaking for the UK, every stroke center will want one. Dirt cheap as well. [1]https://news.ycombinator.com/item?id=22340458


> This would still be hugely helpful for brain imaging en route to a medical facility

I'm speculating, but probably not. The images would be blurry and unusable if taken in a busy ambulance on road to the hospital. Could do more harm than good.

Having this in a stable home would be awesome on the other hand, but there is too much vibration and everybody trying to grasp handlebars and keep equilibrium when you are in an ambulance on duty.

I could be wrong. I do not master this technology, but if you need absolutely remain motionless in the big machines, I assume than smaller ones would have the same requisite. Ambulances are not stable by default. Everything bounce.


MRI is not an emergency technology. Most hospitals will outright refuse to do them for the vast majority of inpatient stays.

This technology will mostly end up in imaging centers, improving throughput.


When I slipped on some ice eight years ago and hit my head, the ER gave an MRI. Now I wonder if they were just pumping up the bill, which would be strange because at the time they thought I didn't have insurance.


Could be pumping up billings, but odds are decent they weren’t. An insurance company’s default would be to refuse payment on an ER MRI, so the docs must have felt reasonably confident they needed it and could justify it to insurance or were willing to get reamed out by management if they couldn’t.

Still, an ER MRI is wildly unlikely. “Possible evolving emergency” and “an hour in a tube where we can’t see you” don’t play well together. Any chance you’re confusing CT scan for MRI? Cause -that- we would do for a head trauma like 1000% of the time.


I had an MRI at the ER for a paralysis attack several years ago. I stabilized after IV fluids were given, they found extremely high neutrophils and thought sepsis, but I was lucid and had no fever. They gave me the option of waiting there for ~6 hours for the MRI shift to start, and I agreed. they found no sepsis or cauda equina, discharged me and told me to come back if it happened again. fortunately, it hasn't yet.

I think the combination of weird blood work + medical mystery + stabilized meant I got an MRI. wish they'd gotten to the bottom of it..


You can do useful imaging in well under 5 minutes with modern MR, so it’s not like like this is fundamentally impractical, it the workflow would have to be tightened up.


Is that 5 minutes with a strong field of a conventional machine?


Yes with acceleration techniques, etc. Main field strength isn’t the real issue here it’s how quickly you can set up the other fields repeatedly and how long your pulse sequences run. There are a lot of factors, but somethings can be done quite quickly.


Noooooope lol


This is a bit of a chicken and egg problem. Not sure the units referenced will shift it, bit a good part of the reason MR is not used for emergency workflows is related to siting issues and cost (hence workload, you can’t afford to have one sit idle easily ) as well as time and complexity of imaging.

If you solve or alleviate those issues it becomes more attractive. There is a fair bit of research around potential trial benefits if the contrast provided for some emergency procedures already.


It is used in differentiating strokes vs aneurisms and extremely important in emergency care for that reason.


Sort of? Not quite?

Ugh, I just wrote this long-ass post and lost it. Whatever, here's the short version:

Stroke is a clinical diagnosis, not a radiologic one.

CT will catch hemorrhages or large infarcts just fine. That's the important stuff: big infarcts cause big tissue damage. Tiny infarcts cause tiny tissue damage. MRI's superiority over CT is in that it catches smaller infarcts, and catches infarcts early. If I see a patient that screams "stroke" and the noncontrast CT is negative for hemorrhage, it's ischemic until proven otherwise. I don't need an MRI at the less than six hour stage and, while having one would be nice for malpractice reasons, whatever, I don't actually need one for your care. I may follow-up with an MRI if it's less certain that it's a stroke: tumors, infection, etc. that look like strokes. But, key - those aren't pressing, time-wise, the way a stroke is, if they look like a stroke ("look like" implies things about the nature of the underlying condition.)

After 6 hours we've passed the TPA stage and we're looking at clot removals. Here we need to know the exact territory of the infarct. If I'm bothering with imaging, yes, MRI/MRA, but that's not about finding the kind of stroke - it's about telling the interventionalist where to go.


(Kind of unhelpful after the fact - but regarding chewed posts:

The method that works where I didn't accidentally close the last tab associated with a particular renderer process (or, TL;DR, didn't close any tabs), is to promptly figure out which renderer process is behind the tab in question via chrome's task manager, quickly attach to it with gdb, then `generate-core-file (filename)`, and grep through the result for my text. (Usually `grep -aC100`.)

Alternatively, hibernating generally seems to always(?) save the disregarded bytes to disk (the associated memory regions immediately get used for something else, so they stay in in the pagemap); then I can reboot the machine, carefully prevent it from resuming, boot to Linux, mount the FS readonly as appropriate (since it was never unmounted cleanly), then go through the hibernation/swap image. (Bonus points for then resuming once you're done.) I recovered a family member's lost draft email this way once (TIL Gmail encodes emails in both HTML _and_ HTML-inside-JSON O.o).)


It's used in specific emergencies: https://news.ycombinator.com/item?id=22340458


Rather than emergency situations, I imagine this would benefit hospitals in poorer, remote areas. For example, an island hospital which have no access to adequate equipment to give adequate diagnosis. Islanders end up having to be airlifted or transported by boats to an inland hospital for anything semi-serious and above.


> The magnet is low-power enough that you could walk within a few feet of it during operation without danger.

That was going to be my question. From my current understanding you can't go into the same room as an MRI scanner with any metal in your pockets. The portable nature of this device made me question the safety aspect.


I think the real takeaway is that this kind of device is possible and they've given a blueprint. It will now be possible to start engineering new form factors, have scanners that are mobile, or maybe scan dogs and cats.


“According to Connecticut-based Hyperfine, their machine will cost $50,000, which is 20-times cheaper than traditional systems, runs on 35-times less power and weights 10 times less than normal 1.5T MRI machines.”

I believe GE came out with MRI machines a while back that were specific for limbs. This really does not seem much different other than it is on wheels and you can possibly position over a torso in a bed if it isn’t metal.


A hospital billed my insurance $15k for a single MRI series once. This machine could pay for itself very quickly.


Lots of hospitals love to use MRI’s as a high margin business. I had a contrast MRI where my insurer was billed $20k and paid $16k.

Same MRI at a stand alone clinic is about $2k.

It’s ridiculous.


A 1.5T torso MRI costs me $260 US dollars in Mexico without insurance and without prescription, radiologist interpretation included. Salaries are higher in the US and I assume the bleeding edge tech is multiple millions of dollars. But even with those considerations, I don't think $20K is justified.

Sure, they don't have fast 3T machines, but still are useful for DIY cancer screening for the paranoid like me (for those that aren't familiar with imaging tech, MRIs don't use radiation).


In Russia torso MRI costs around $200-$300 too. Smaller exams on outdated machines could be found for as cheap as $80.

It is solely US thing that medical procedures have astronomical cost.


In Italy, my mother did a few months ago a contrast MRI (spine), directly paid in a private laboratory (outside the national service) for around 300 Euro.

Which makes sense if the machine cost is around 1,000,000, appointments were every half an hour, so, let's say they make 15 scans per day at 300 each, it makes 4500 Euro/day by 220 days/year that is around one million Euro/year.

You take out of it - say:

250,000 Euro for the machine (for 5 years)

250,000 Euro for personnel

250,000 Euro for maintenance, power, other costs

50,000 Euro for something else I don't know

And you are still ahead for 200,000 Euro/year, i.e. roughly 20% of your investment.


Brings up a point never have imagining or blood work done by a hospital if you can at all avoid it.


Especially if you keep billing insurance $15k to use it.


There is a remarkable asymmetry between physicians who take risk to read the mri and are paid a few hundred dollars and the corporate stooges who run the hospital and charge whatever they like. And yet with such a willing supply of doctors from abroad there is no end in sight.


Small models used to scan lab rats also exist (a friend does research on brain damage prevention and I was wondering if they let her use human-rated machines for murine models).


The small-bore machines they use for rodents have much stronger fields: 9.4 and 11T field strengths are pretty common. This is a lot easier to achieve in a small bore. For comparison, the “standard” research scanners for humans are usually 3T, with 7T scanners becoming somewhat more available.

That said, many universities do let people use the human scanners for other research. It’s kind of a pain to do with animals, because you’re paying $500 an hour while you scrub the room down afterward and wait for the air to cycle, but it’s the only way to scan larger animals (primates, pigs, dogs, etc).


Really nice piece of work.

The key patent makes for interesting reading [1]. They use neodymium permanent magnetic rings. I wonder if building a crude version of Hyperfine’s machine is within the realms of a DIY project.

[1] http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=H...


The MGH Martinos Center has a very nice open source MRI page with hardware and software instructions to build a functional, low-resolution tabletop MRI (I think this is primarily targeting instructional use at a university level, so requires a fair amountof background):

- https://www.opensourceimaging.org/project/instructional-tabl... - https://tabletop.martinos.org/index.php/Main_Page


This is great. There is a lot of work being done to figure out how much undersampling you can do on a typical MRI and reconstruct a good image. For typical scanners, this gets you faster scan times. It will be interesting to see how much those techniques translate into taking a super low-powered MRI and allow it to yield decent images.

Whether or not such techniques are good for accurately detecting lesions seems to still be a bit of an open question. (It's less relevant to my particular research interests for now, but obviously extremely important for clinical use.)

A lot of the fallout from over-imaging (especially when, like MRI, there is no ionizing radiation so the test itself isn't harmful) is due to physician behavior and standards of care. But those behaviors and standards can change over time.

A bit tangential, but I suspect that at some point, we will think it anachronistic that cardiologists tried to visualize the heart's function by listening to its sounds with their ears.


"To reserve your Hyperfine system(s), the following conditions apply:

    You must be an accredited US health care provider.
    System usage is approved for patients 2 years and older.
    Your institution must be able to supply a wired ethernet connection to the device once every 14 days."


Curious about that last condition. What is it used for? And what bandwidth does it need?


I'm only familiar with other, not medical, devices and my guess would be that this is about firmware updates.


Licensing and/or updates?


I heard that ordinary MRI machines can't ever be powered off or they heat up and the expensive liquid helium boils off. How does this particular machine overcome this issue? Seems like it would be a huge problem for a portable MRI


High field magnets are superconducting alloys that must be kept under liquid helium to remain below their critical temperature. If their temperature rises above the critical temperature, they become Ohmic resistors and can melt -- or at least boil off all their cryogenic helium and nitrogen rapidly enough to be dangerous (so called quenching).

Low field magnets are usually just solenoid coils. There are a number of low-field magnetic resonance devices, including benchtop relaxometers, which use either permanent magnets or electromagnets. This one just uses an electromagnet, I believe, and it looks like it's in a Helmholtz configuration. This is a smaller version of a design used primarily in veterinary contexts, e.g. so you can walk an upright horse into it.

(Edit: Someone below says it's permanent magnets here. Can probably look up their patent or FDA application to be sure.)


I think fixed machines are like 1.4 - 3 Tesla field strength. The portable one is like .05 Tesla... So maybe it doesn't even use superconducting magnets.

But you can get high temperature REBCO tape now and get super conductors that can be cooled with liquid nitrogen.


MRI machines go up to 7T now.


Wow, makes me wonder about adverse effects. Could you magnetize the hemoglobin in your blood?


.05 Tesla! Seems like there would be poor contrast and SNR. I need images.


I understand that there are now graphene based magnetic field sensors which offer orders of magnitude improvements in sensitivity, from the likes of Paragraf. I'd be surprised if they had already made it into product but low intensity MRI is certainly an application touted.


MRI/NMR doesn’t sense magnetic fields, so I don’t think that kind of sensor would help. You have to put your sample in a strong magnetic field for this method to work, but after that it’s just a matter of emitting some RF pulses and receiving some RF signals.


I thought they had to turn the magnet on and off. This would first align all the magnetic moments of the specific atoms and then as they randomize when the field is off, they interfere with a superimposed radio field creating peaks and troughs. Tomographic calculations are then applied to the collected radio signals to give an image. How can a permanent magnet be used?


There are a lot of NMR imaging techniques, and I'm not an expert, but my understanding (based on discussion with experts) is that the main magnetic field is static during all MRI scans. The gradient fields from electromagnets change, though, to alter the Larmor frequency across the thing being imaged.


The field is always on. Safety first. :)


If I understand it correctly, the machine uses permanent magnets rather than a giant solenoid.

The magnetic field is way less strong (0.064 Tesla compared to at least 1 in most machines), but the machine doesn't have to generate heat while producing a magnetic field


An MRI magnet doesn’t produce heat while producing a magnetic field. That’s not why they are cooled. They are superconductors, the resistance is 0.


> They are superconductors, the resistance is 0.

So long as they stay under their critical temperature. If that changes… well, a lot of liquid helium is about to stop being liquid.

And that's one of the reasons why a permanent-magnet MRI is so interesting -- it doesn't depend on a constant supply of power (for the helium cooler) and gaseous helium (to replace lost gas), so the system is portable, and maintenance costs are likely to be much lower.


I have some experience with NMR from a chemistry perspective. The magnetic field is generated through a super conductive coil. Cool it down with liquid nitrogen, generate the current and you have a superconducting magnetic field as long as you keep it cold.

If it warms up, it’s no longer superconducting and all that current starts generating heat which boils the liquid nitrogen off, which can be quite spectacular. Plenty of videos on YouTube.

So yes, turning the magnet on and off is not trivial at all, but as long as you keep it cold, you’re good.


The ones in hospitals normally use liquid helium, not liquid nitrogen, which accounts for the quench being so spectacular and expensive. The nitrogen just keeps the helium from boiling off too fast.


There is a big red button to quench the magnet in emergencies, but it can cost a lot of money to get the MRI going again.


Money and time. I recall one clinical machine that was still not operational I think something like 15 months after unintentional quench. That’s expensive real estate to be sitting idle.


15 months is exceptionally long. With a decent support contract, it can all be sorted out in a few weeks.


In theory it can, but depends a bit of damage and the state of your support . I suspect that one was due to arguing about the latter than the former.


There are superconductors that can use cheap liquid nitrogen now.


I was talking recently with a Haitian doctor, who said that in his entire country (of 11 million people), there is just one MRI machine.

This new device seems like it will break down so many barriers, both in terms of cost and practicality, in Haiti and other developing countries.

I hope it makes its way there!


It's a ridiculous thought, but I couldn't help but think of the "Thou Shalt Not Covet Thy Neighbor's Goods"[1] skit from "The Ten" where neighbours keep one-upping each other by buying CAT scan machines. It's really crazy to think you could almost have an MRI in your own home these days.

[1] https://en.wikipedia.org/wiki/The_Ten#Plot


Reduced power and cost requirements are critical for this technology to be viable for low middle income country (LMIC) settings, particularly in rural areas. In many sub-Saharan countries, there are A handful in the country and use is complicated by unstable grids. High impact tech if use cases for these settings are sharpened.


From a quick survey of the article, it sounds like this instrument is likely to be using conventional electromagnets. Is that correct?


Permanent magnets, according to this article:

https://www.fiercebiotech.com/medtech/fda-clears-world-s-fir...

Small permanent magnet NMR spectroscopy units first showed up in chemistry about 10 years ago:

https://www.prweb.com/releases/2010/11/prweb4732684.htm


Ah. Cool. Makes sense, as the field is clean and stable. NIST's Kibble balance does the same. Thanks!


My understanding is that MRI’s use two magnetic fields in opposition. A strong field that is a superconducting magnet of fixed number of Teslas, and an electromagnet that is pulsed to cancel out part of the field. The pulsing creates an echo in the water molecules in the body which is then painstakingly processed into an image. That thunking noise has to do with the pulsing.

I forget what it’s called now but you can build a permanent magnet of upward if .5 or 1T by packing a bunch of keystone-shaped rare earth magnets with differing field alignments into a high tensile strength retaining ring. This causes the fields to be additive, but God help you if you manage to break the damn thing. I watched a video on how to safely put two large rare earth magnets together without shattering them, and YouTube recommended an infomercial by some company at the high end of Tesla ratings for permanent magnets. It was kinda nuts.


Portable is a relative term.

Some hospitals in Germany actually have vehicles containing MRI units which they dispatch when a stroke is suspected.


NFL gonna buy one of these per team


How does MRI compare to other techniques (e.g. CAT scans) for general diagnostics? Does it make sense to be used as a more routine diagnostics tool if there are many more machines available?

In the US there are 37 MRI machines per million population, even in other relatively rich countries like France this is much lower at 14 per million population [0]. Is it just the cost of the machines, or are there drawbacks that limit their use as general purpose diagnostics tools?

[0] https://www.statista.com/statistics/282401/density-of-magnet...


It would reduce their worker’s compensation insurance costs.


I don't work in medical devices per se, but I do work in hardware.

I see lots of immediate uses for this. Pro-sports will obviously want it, the military will too, field hospitals in e.g. Africa. Lots of possibilities here.

Can anyone tell me where this device will be manufactured?


Ten years ago I got an MRI when they were searching for kidney stone. I think the hospital charge was over $3000. I had one last year and the charge was less than $500. You can point to MRI as one of the success stories for technology in health care.


I paid $2000 for a wrist MRI scan in US and $200 for one in Nairobi.

Basically anything in US will be 10X expensive because we have wonderful and very efficient healthcare for doing even the most basic things like CT, XRay, MRI etc. /s


Hospitals around here charge ~$2000 for a CT scan. The price isn't coupled to the cost of the machine.


A computed tomography scan is not MRI. It involves ionizing radiation (xrays). Not that an xray scan should cost $2000.


Yes, I'm aware of the difference. Typically, CT is more available and cheaper, which is why I thought it was a relevant point of comparison.


Which country? Doesn’t sound like a US hospital.


You can get a world class MRI in the Bay Area for $550 from SimonMed Imaging.


You can also get a worse MRI on obsolete equipment at UCSF for $18k. Ask me how I know!


Is that cash price? Does that include the radiologist interpretation? And can you get one without a prescription?


Cash price. Based on how the thing was billed, I am 90% sure it included interpretation, but at this point I can't remember for sure. I don't know whether the prescription is mandatory, but I'm sure you can email/call them.


Now we just need a good AI to read it.

This would be especially useful in the developing world.


Radiologists are cheap and plentiful in the developing world, compared to MRI machines.



Fair rebuttal, but I don't know if medical cartels would allow AI to replace their limited supply of cardiologists so easily.

It is quite cheap here in India already (got an orbital MRI last week for $105.10, including radiologist's fees; it can go as low as $40) but having it cheaper and more easily available would help a lot more people.


There's a ton of small clinics around the U.S. unaffiliated with hospitals that offer MRIs for far less. Just Google it. These places will usually even give you a price ahead of time, you'll never get that from a hospital.


Remarkable they have enough sensitivity at 0.064T. To me, that’s the underlying scientific breakthrough. It implies that that there are other, intermediate regimes that would not require bulky, expensive superconducting magnet.


I wonder if it would be possible to up sample the images using AI/ML?


I wonder how long it’ll take for these to appear at the airport checkpoints.


I wonder what the resolution on these will be like. As I understand it, bigger magnets and more power = higher resolution.

Will this thing have enough to be useful?


The reason for huge magnets is to drown out effects of external magnetic fields which would otherwise cause spatial uncertainty. For me this seems like something that can be partially compensated for in software postprocessing given fast enough scan rate.


The magnetic field is to shift the energy levels of the nuclear spin states. The transition energy is proportional to the applied field, and the wavelength of the RF field used for imaging is inversely proportional to the energy. So a stronger field gives a smaller wavelength and thus higher resolution.


Oooh, in the same way that the higher 5G frequencies don't travel as far!


This is really not correct. External fields are tiny relative to these fields, you have bigger issues with coupling and eddy currents.

It’s not quite as simple as field strength equals resolution, but generally it is easier to achieve high resolution with higher main field strength. See for example mouse scanners at 12T




Consider applying for YC's W25 batch! Applications are open till Nov 12.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: