Yet the press this is getting is, like the article said, a black mark on Philips. They were a very diverse company, until they chose a few specific markets in which not to divest, healthcare being one of them. This must be quite the chill at headquarters.
The fact that nobody was injured in the previous incident is fortunate for anybody who might have been injured then, but doesn't diminish the importance of addressing whatever flaw caused the explosion. Considering that we're talking about a lot of liquid helium being improperly vented, the next time something goes wrong it could end up asphyxiating a hundred people.
Just because nobody was in the danger zone the first time it exploded doesn't mean you should count on nobody being in the danger zone the second or third time it might explode. This is the reason work site safety officers yell at people to report all near-miss incidents, not only the accidents that get people hurt. If something went wrong and nobody got hurt by chance coincidence, it might happen again when people are actually near it. Therefore a near miss incident needs to be addressed with the same seriousness as an accident that actually caused injury.
Even if it doesn't suffocate you, helium leaks can be pretty bad since they'll temporarily brick any modern phone with a MEMS oscillator (like a quartz time crystal, but smaller) in it. Not great if you need assistance.
That's absolutely a recall. An FDA recall means the product is in violation of the law. Those violations must be corrected or the device must be destroyed, and if not, then the FDA has the right to seize it.
The recall strategy in this case is not to ship it back to the manufacturer, but to have the manufacturer come out and service the device until it is again compliant with the law.
It has very little to do with the violations of the law. There is no law that states mri machines must have <X% chance of exploding.
The way these things work is that the firm has decided based on their data and reports to issue a recall and notify the FDA. The manufacturer recommends hospitals not use the device, however there is the possibility that they continue to do so (the probably won't for liability reasons).
I do medical risk assessments professionally and this is not at all how it works at all.
Most FDA approved medical products have inherent risk, e.g. X% of people will die.
The question is actually about how the actual performance aligns with disclosed and approved risks.
If Y% of people are hurt, the expectation was X%, and Y>X, then this is grounds for a penitential investigation and potential recall.
If it actually worked the way you stated, these machines would probably never have been recalled.
I dont understand what the fault is - quenching is bad but happens. Blocking the quench path sounds bad - is it saying that something is causing it to be blocked or is the machine supposed to do something else if hte quench path is blocked (what can it do??)
The rapid conversion from liquid or solid to gas, and the means to (briefly) contain that... is a bomb. The reason you don't block the emergency relief is that you turn a quench from an emergency involving asphyxiation, to something resembling a detonation.
It's a certification requirement that the magnet can quench with a patient inside and the patient is not affected – the cryogens are supposed to vent outside the room, or, in the even that does not happen, at least safely within the room.
A blocked quench pipe and a stainless steel (typically A4 stainless) cryostat becomes a bomb: cryogens expand in volume by about a factor of ~700 when boiling and so the volume they occupy at 1 bar and 300 K is, well, far far larger than the volume of the room the scanner is in.
This is the sort of thing that is a (remote) risk to life. I've only ever seen one (preclinical, not-for-human-use) magnet unexpectedly quench; copying and pasting a paragraph from my PhD thesis:
*The overpressure created by the rapid boiling of the large volume of liquid helium surrounding the superconducting magnet was sufficient to burst the bursting disc (as designed), which allowed the venting of helium gas from the inner cryostat to the emergency quench pipe (the bursting disc was temporarily replaced by clingfilm in this photograph). Additionally, it was also sufficient to rip and blow off the quench pipe, break two windows, damage the floor, and relocate several light items within the magnet room (which, presumably, was not as designed)*
"During a quench, which is not common, a large amount of helium evaporates and is vented outside the building through a venting system," the recall explains. "If an unknown blockage is present in the venting system and the pressure exceeds design limits, the structural integrity of the system could be compromised."
This is a risk of any mri machine that uses helium to cool a magnet. Why are only these machines being recalled when if you block the exhaust of any mri scanner it will blow up?
This is just speculation, but maybe there is a secondary rupture disk that vents the helium inside the room when the primary path (the exhaust to the outside) is blocked. And maybe that secondary rupture disk is too strong, so another random part of the helium containment fails, blowing up the machine and possibly injuring the patient. But please note again that this is pure speculation.
Why not add a pressure relief valve on the quench path with a very loud whistle? That should be enough to take care of such rare and compounded failures.
What does recall means in this context? De-energizing the superconductor and shipping it back? Seems like a waste and a planning nightmare.
A bursting disc is commonly used -- the diameter of the quench pipe is typically around 20 - 30 cm. The gas flow rates are insane; a PRV would fail and likely still not reduce the pressure inside quickly enough.
Remember, cryostats are like Russian dolls suspended on torsion wire. You want the mass of the metal inside to be as low as possible because it forms cold bridges to the outside world and increases the boil-off rate. Quenches should not happen once the magnet leaves the factory, but until that point it's not uncommon for a machine to have several "training" quenches as the (typically NbSn or NbTi) superconducting wire effectively anneals in place. A fixable giant hole in the top (with a graphite, insulating series of bursting discs) is the approach usually taken.
I do not work with MRIs, but I work next to the guys who run the NMRs (which is the ~same technology). It is my understanding that all of these super cooled magnets are designed for the eventuality of an emergency quench. Which means the machine has a direct path to evacuate the gas, and it should be piped into the building's HVAC so that if a quench does happen, the people in the area do not suffocate because of lack of oxygen.
A surprising amount of maintenance can occur while the magnets are cold and energized. My armchair-uniformed-guess is that they can replace the not-always-working relief path without venting.
The problem is you dont know what you are talking about and dont even understand the questions people are asking.
Every MRI machine quenches and vents. The question is what is different, in detail, that heightens the risk with this machine.
your response is like saying "the engine wont work" as a complete explanation for why a type of car has more automotive problems than another. Then, when a mechanic is curious about the nature of the failure, you call them stupid, saying "I already told you, i said it doesnt work!"
I certainly hope you are not in charge of solving problems in any professional capacity.
> Every MRI machine quenches and vents. The question is what is different, in detail, that heightens the risk with this machine.
The design can't safely accommodate the level of energy that ends up in the machine if its clogged.
In the spirit of analogies, it's like asking why a 10 oz glass overflows with 200 degree water and a 10 oz glass doesn't overflow with 100 degree water. It exceeded it's capacity.
There's nothing more complicated to it, you could call in an MRI technician, get the PDF manual, talk to several doctors, disassemble it and put it back together, and learn no more than that.
You got extraordinarily rude and personal, I hope you don't do that often on HN, that was an outlier in my 13 years here. I hope all is well and it's an exception.
The Helium used for cooling expands a lot going from liquid to gas during a quench (iirc by a factor of ~1800). The pressure built-up from the phase change has to be relieved *quickly* before the machine turns into a crude pipe bomb. Oh and you can't get enough Oxygen from the air if most of it has been replaced with leaked Helium.
Note: MRIs are incredibly safe. Your risk of dying due to an MRI machine exploding is significantly less than your risk of dying due to falling while climbing into the machine.
You should respect the magnet though. Those warnings aren't there without reason, getting between an MRI or NMR and a magnetic piece of metal is a bad idea
Oh, absolutely. MRI machines are incredibly dangerous if not handled properly. But the overwhelming majority of MRI-related dangers are well understood and avoidable.
~~I don't know why this is relevant, and quite frankly I find comment odd. Just because y is safer than z, does not mean we should ignore reasons why z isn't safe.~~
See OP's response for why this is crossed out (assuming HN supports it).
I think it's an important point to reinforce because news stories like this can make patients afraid of MRIs, potentially resulting in them avoiding necessary imaging. Most people on HN are probably not going to make that error -- but I think we have a responsibility to help communicate to the broader community in this regard.
At what point are less safe but cheaper machines worth it. Hospitals ration MRI machine access through out the world. Even in Canada, you have to wait on a list to get access.
"Canada could expect to wait a median of 5.4 weeks for a CT scan and 10.6 weeks for an MRI scan"
thats because Canada has less than 20% the MRI machines proportionally to population as Japan, and 25% of the MRI machines as the US. We're down near the Czech Republic and Turkey.
FWIW in the UK I was able to get a MRI the next day when going private. They even gave me a CD with the images.
They would have even done it the same day but I wasn't available.
I suspect in the NHS I wouldn't have even got a MRI at all (it was for a mildish running injury to my knee and just needed some physio in the end, so MRI seemed like overkill!)
I worked in x ray imaging software in a company that assembled them.
Risk assessment and traceability are not fun, but it's interesting to see how to make safe things.
I wish general software had similar constraints or designs, or at least insurance companies would lobby the government to force software companies to proof read their code.
I don't want rust everywhere, but it should still be possible too approach what rust is doing by other means.
Toxic teams and lazy developers will also find way to produce crappy software with Rust.
The Rust developers of today might produce better software, because they are the early adopters who care about performance, safety and correctness, but I believe (assuming Rust becomes mainstream) if you give regular developers Rust to code in, they will produce the same quality as they do with Java, Javascript, or C.
The environment and the individuals are significantly more important than the programming language they use.
This is about quenching, which happens in emergencies or failure. During regular use helium is recycled in similar systems often today. But it's simply not feasible to do that when all the liquid helium in there evaporates at once
This only happens in an emergency, not during normal operation.
Pretty much the only time you would do it when a person is stuck in the machine by some piece of metal that is attracted to the magnet. You hit the quench button which vents all the helium, but more importantly also halts the magnet (all the energy released from halting the magnet goes into boiling the helium), and then you can release the person.
You don't do this for any other purpose, for example if some metal cart is stuck in the machine, but no one is at risk, you use a slower method to shut down the magnet (i.e. drain the energy) that doesn't vent the Helium.
I was under the impression that metal 'sticks' to an MRI machine with so much force it will usually puncture the machine and all the helium will leak out ..
The issue in that respect is that they flood the coils in a helium bath, instead of utilizing additional structures like channels/nozzles to provide the same rate of freshly cooled helium to the heat sources as the bath does, but with far less helium required in the system.
I don't think any similar magnet exists that doesn't embed the magnet in a dewar full of liquid helium. And it probably doesn't make any sense to do it differently, even if it were possible.
You need some mass here for the helium anyway as you never ever must run dry and it constantly evaporates.
Its unbelievable the decline of philips, i used to buy their stuff out of trust, but given their story with ventilators, this one, and other personal ones with razors, waterfloss and toothbrushes, I wouldn’t touch anything from them with a ten foot pole
> The lawsuits have claimed that flaking foam and gasses emitted from the machines were linked to health issues including respiratory illnesses, lung cancer and death. The foam was used in the machines to reduce noise and vibration.
> In June 2021, the Food and Drug Administration announced a recall of Philips machines that also included BiPAP devices and ventilators made since 2009, warning that foam deterioration in the products could cause “serious injury” to users. Philips initially released a memo to doctors saying the foam breakdown posed risks of “toxic carcinogenic effects,” but the company has since released updates reporting a far lower level of concern.
> So they don't explode when they're new, but after a while? How old was the machine exploding?
No. What they say that the problem occurs only very rarely. The first of this MRI model was introduced in 2001, there are hundreds of them used all the time and only one explosion happened. Presumably that is the explosion which has shown them that there is something wrong with the model, and now they are taking corrective action.
> this would be the second recall in few years for philips
It is a huge company doing a lot of things. The safest way to never make a mistake is by sitting on your hand and refusing to do anything.
Some of the affected MRIs are old enough now to buy you a alcohol, how is this an evidence of anything recent with Philips?
> whatever, doesn't come through as high quality
They found a rare but potentially high impact issue and are fixing it in decades old equipment on their own dime.
Show us that you have done things of similar complexity better. What equipment you have designed is in constant operation since 22 years? How do you know it doesn't have any low probability, high consequence failure modes?
Maybe if they do too many things that need to be recalled they should cut somewhere, I don’t do MRIs that’s why I dont sell/recall them, making things personal on this scale seems not very smart argument btw
> making things personal on this scale seems not very smart argument
I think it is very illuminating. Imagine you created a thing. It is successful enough that hundreds of hospitals bought it. They use it day after day for healing people. And after 22 long years ago one of the many many decisions encapsulated in that product is a mistake in hindsight. You worry it might cause harm in rare cases. But, luckily you can send around a service technician who checks things, maybe takes corrective action and it further lowers the already low chance of anyone getting hurt.
Would you feel like a failure? I would be over the moon that I made something so successful and helped heal so many people. Would be slightly embarrassed that I have missed something, but glad that nobody got hurt and we can fix the issue.
If this is how I personally would feel why should we act as if the company as a whole is a complete failure?
> Maybe if they do too many things that need to be recalled they should cut somewhere
You don't understand me. It is not that they do too many things and that is why they have recalls. Every human endeavour has the chance of things going slightly wrong. If an entity, a company in this case, has many projects it is more likely that you will hear about some of them being occasionally going a little bit wrong.
What I'm saying is that you should look at the number of recalls as a proportion of their reach and volume. And if you do that you will realise that they are neither exceptional nor should they feel ashamed of themselves.
It sounds like you demand perfection and that is not a realistic goal to attain.
(edit: actually not a recall but a requirement that the machines are not used until they’re serviced)
https://www.fda.gov/medical-devices/medical-device-recalls/p...