Comment by Someone1234 | Hacker News Reader

Comment by Someone1234 | Hacker News Reader

This is definitely interesting, scientifically, and may help us better understand Autism (and its causes).

I cannot imagine this will see widespread adoption outside of research. MRI scans remain expensive and many MRIs have waiting lists at hospitals. Even today, they regularly use less safe technology (e.g. CTs) because they're less expensive and there will be more available.

There's no safety reason why you shouldn't do this. MRIs are darn near harmless. It is just an economic issue with gaining time on an MRI to discover information which may not change the overall outcome for the child (pediatricians are getting better and better at detecting early signs of autism).

Heck, if MRIs were readily available they'd quickly replace ultrasound during pregnancy (since MRIs are likely safer and superior given that they better represent depth). But alas, a MRI costs up to $1m for a "basic" one and up to $3m for a fMRI, and that's ignoring the space/training/power utilisation issues.

I've been working on a bootstrapped company building a low-cost MRI for about a year now. We're hoping to have a demonstrable prototype in another year or so.

So we think that we can get to that future where people pick MRI based on application and not cost or availability.

I'm so glad to hear that. I've been thinking for years that MRIs seem like an industry that could and should be disrupted. The benefits could be massive.

I'm not in the medical field, but I really believe the future of cancer medicine will strongly lean towards prevention and early detection, as opposed to just cure.

Regarding cancer, we think you're right. That and possibly maintenance - for example, most prostate cancers are asymptotic for life if untreated. For slower cancers that are acquired in old age it can make sense to use less aggressive chemotherapies that maintain quality of life. We think that being able to monitor the development of the cancer closely and frequently is a key component of enabling future therapeutic approaches like that.

Where I live there used to be radio advertisements for a service that would give you a full body MRI in an effort to find cancers that could be treated before they spead.

After a bit of reseach I discovered it's a false promise. A lot of people have benign lumps and growths they can carry around for a lifetime without any trouble, and while MRIs are quite safe biopsies are not. If you don't have a reason to think you might have cancer, i.e. the MRI is a screening tool and not being used for a diagnosis, the very small chance you'll die of complications from unnecessary biopsies outweighs the very small chance you'll save your life by discovering a treatable malignant growth.

At least, that was the state of the art 7-8 years ago.

I've heard the same said of getting mammograms started earlier in life: they don't recommend it because the time spent chasing benign lumps is neither safe nor efficient.

This always seemed to me a terrible indictment of how medicine is practiced. Knowing that, do you have to chase down every lump your screening finds? Can you not consider the benefits of investigating further based on the dangers of doing so, the risk factors the patient has, maybe even what you see in successive scans over time?

I am not a doctor, but there seems something unhealthy about a system in which you are advised to avoid learning information about yourself in order to avoid treatment.

I think you're exactly right - more frequent, less expensive testing means we have to change the criteria that mandates further investigation.

One dilemma I've been thinking about recently is that there's an unfortunate burden put on care providers due to litigation that results in increased cost - the provider or radiologist that sees a "maybe" or a "shadow" on an image has to order a biopsy just to protect themselves from being sued on the off 1% chance that the thing becomes a serious health concern. In an environment where consumers evaluated their medicine based on value/$ they would be further dissuaded from a biopsy, but in America the nature of medical insurance disincentives this sort of cost analysis skepticism.

The incentives aren't aligned. A doctor who sees a lump and does nothing has a huge downside risk: it could be malignant and everyone will believe that the doctor is at fault. A doctor who orders the test will never be blamed for a fluke complication during a biopsy.

It's hard to resolve becuse it's not even about money. It may be the doctor who is blaming himself.

The doctor or the patient; if you're told that there is a lump (or seventeen) and it's probably nothing but you can get it checked out if you really want, most people will blame themselves if they don't get something checked out that later turns out to be something, and they know it.

That doesn't make sense to me. I have, many times in the last year, had a doctor say to me some variation on, "this isn't dangerous, but it could become so -- we'll keep an eye on it."

So it's not that they can't act with sense and restraint with respect to risk. They clearly can.

With respect to breast cancer specifically, they told me in school that finding it early improved your odds, and the earlier the better, and if you really wanted to do the right thing, you'd physically examine yourself looking for lumps on a monthly basis. That didn't seem unreasonable - and yet when something more accurate and objective is available, it's to be avoided out of a fear of unnecessary biopsies?

Something strange is going on.

> you'd physically examine yourself looking for lumps on a monthly basis.

That's not the current advice. Current advice is that women become aware of what's normal for their breasts, and look for certain specific changes and not the vaguely described "lumps".

http://www.nhs.uk/chq/pages/1740.aspx?CategoryID=60&SubCateg...

> and yet when something more accurate and objective is available

Is it more accurate? Mammograms are sensitive, but have a high false positive rate, especially for younger women.

(All the sites I can find report sensitivity and false positive rates as percentages, which is a terrible way to report this information.)

https://ww5.komen.org/BreastCancer/AccuracyofMammograms.html

> Overall, the sensitivity of mammography is about 84 percent [9]. This means mammography correctly identifies about 84 percent of women who truly have breast cancer.

> The more mammograms a woman has, the more likely she will have a false positive result that will require follow-up tests. The chance of having a false positive result after 10 yearly mammograms is about 50-60 percent [22-24].

http://www.breastcancer.org/symptoms/understand_bc/statistic...

> In 2017, an estimated 255,180 new cases of invasive breast cancer are expected to be diagnosed in women in the U.S., along with 63,410 new cases of non-invasive (in situ) breast cancer.

Ann, a 45 year old woman, goes for a mammogram. It returns a positive result. What are the chances that Ann has breast cancer? (Scarily lots of doctors can't answer this question if the information is presented as percentages. See Gerd Gigerenzer's book Reckoning with risk for many examples.)

https://www.cancer.gov/types/breast/hp/breast-screening-pdq

> Magnitude of Effect: In the randomized controlled trials (RCTs), for women aged 40 to 74 years, screening with mammography has been associated with a 15% to 20% relative reduction in mortality from breast cancer.[1] Absolute mortality benefit for women screened annually for 10 years is approximately 1% overall, ranging from 4 per 10,000 women who start screening at age 40 years to 50 per 10,000 women who start at age 50 years.[2] Based on the 25-year follow-up from the Canadian National Breast Screening Study (CNBSS), an RCT of breast cancer screening,[3] there is some uncertainty about the magnitude of benefit of mammography in the present day.

Ah! Thanks!

> Knowing that, do you have to chase down every lump your screening finds? Can you not consider the benefits of investigating further based on the dangers of doing so,

The problem is this: How do you when to investigate further and when not to? Unless you can come up with a way to make that choice without just rolling dice, you're not really doing any better.

> ... the risk factors the patient has, maybe even what you see in successive scans over time?

Aside from perverse incentives as mentioned by someone else, there may also be a risk from the scan itself (as in breast cancer screening, e.g.). There's also secondary effects from a positive scan like anxiety before being able to confirm or reject a diagnosis -- this has known health effects. The confirmation itself may have associated risks, etc.

It's far from clear that the benefits outweigh the risks[1], so it seems like overkill to call this an indictment of medicine.

(I should also add that in e.g. breast cancer women who are known to be at risk -- there's specific genes which carry very high risk (BRCA1, BRCA2) and other genetic factors with high penetrance.)

[1] This is also something that many in the medical profession acknowledge, see for example [2] where there's a load of material on breast cancer, specifically.

[2] http://sciencebasedmedicine.org/tag/mammography/

Getting an MRI and having each (so far symptomless) lump examined is not the goal. But getting an MRI every 6 months to a year and being able to see the changes in these lumps is much more important.

A Skeptical Look at Screening Tests - https://sciencebasedmedicine.org/a-skeptical-look-at-screeni...

So if you got a full body MRI each year/month, you could track the lumps and probably tell the bad from the good.

No, you just generate loads of biopsies. These have risk too, so it's a giant waste of time. MRs produce a lot of data and without knowing what you are looking for you are going to miss things. As with most things, ask a good question, get a good answer.

Really would've thought that with high resolution, time series, and sophisticated algorithms to look at the structure and location relative to what's expected/likely, you could make the data useful. I know what you're saying is true now, but is it inherently true or just a factor of immaturity in the technology?

Immature tech and the way people are. For every lump and bump you see on a person with your eyes, there are a load inside.

We do a lot of research scans on healthy volunteers. You scan certain areas with some trepidation. Scanning the liver of a 50 year old is a great way to find yourself in a world of 'incidental finding' paperwork.

You're missing my point, probably because I didn't make it very clearly.

I'm imagining new technology here.

You'd take the first scan as a baseline. Whatever lumps you have are assumed harmless. Then you track the diffs from there on.

I assume that, at least after collecting experience, we can learn to tell the bad growths from the ignorable.

I remember a similar one with prostate cancer years ago. A significant amount of men will get it but only a fraction of the cancers will develop to be dangerous in a normal lifetime. Removing all cases early on would likely cause more deaths than it prevents.

This is still only an argument to not take surgical action immediately every time something is detected. There should never be incentive to deliberately know less information about your own health.

I've heard this defence before but it is absolutely wrong form a logical perspective. We aren't forced to perform a biopsy on everything that shows up just because we see it.

It seems only logical that occasionally there would be something detected which is unambiguously cancer. The ones that are ambiguous, we could simply monitor without taking action on.

If having more information about the state of your health is putting you in danger, then the medical system is broken. It's as simple as that.

Out of curiosity, what are the cost cutting changes you're making?

I've wondered why there are no MRIs using high temperature super conductors yet, it seems like such an obvious way to cut costs (as you don't need the extreme cryogenics).

I can't go too much into details, but suffice it to say that we're not using super conductors at all.

High T super conductors will be very interesting in the next decade, I think, but right now there are a few issues that I'm aware of:

1) The high T super conductors are almost all crystals, not ductile metals. This means it can be very difficult to form coils or other structures out of them.

2) High T super conductors generally can't withstand as high a magnetic field as the lower T super conductors. I'm not sure how practical a concern this is for MRI.

The benefit of High T super conductors is that you switch from using liquid helium as your cryogen to liquid nitrogen. This is a major simplification, but we're not all the way to room temp yet, unfortunately.

Best of luck! I sincerely hope you're successful.

Thank you!

There several manufacturers of YBCO wire now, e.g. http://www.superpower-inc.com/content/2g-hts-wire

The specs given by the wire you linked is much more impressed than I would have guessed. Particularly the ability to maintain performance at tight radius. Glad to see things are progressing!

Interesting, I'll have to get a quote and run some numbers.

Anything public-facing that you can share with us, so I can follow along over time? I find such work both fascinating & important.

The company is called High Q Imaging. We have a bare bones website that we'll hopefully update in the next month or so: http://www.highqimaging.com/

Thanks!

This will be huge for sports injuries, eg labrum tears. The cost and/or the wait for an MRI is often prohibitive.

Where do you find this to be the case? I'm in California and both times I've had an MRI, I had an appointment within a week (non-emergency) and while I don't remember the cost for the first one, my most recent on last month cost me around $500 - that's not a copay, that's total cost, no insurance.

It's not exactly $20, but I'd hardly call it prohibitive.

Please please please get some radiographer input on the interface. The horror that is out there is just so dire. I have some fantastic screenshots tucked away somewhere of the things that users are subjected to.

Interesting also about the data interface, I wonder whether you can get full data out of a standard MRI, or whether you can only look at it through a proprietary interface.

That area isn't too bad. The format is Dicom and is pretty good at giving you what you want. The format has a ton of metadata per image, and so extra stuff can be written into empty fields for novel imaging applications. Different vendors do different things, but it's not too hard to do a "if Siemens 3T do x" A recent example was perfusing mapping, where different vendors encode data in voxels quite differently, resulting in radically different perfusion maps between vendors prior to correction. It's also not too hard to get the raw data out and actually reconstruct it somewhere else. However the raw data is huge and you would have to really loath yourself to want to do that.

Thanks for taking the time to respond. I've been wondering about this for a while.

If you want a sample file it shouldn't be too hard to get one, but I believe there are various objects that off them - conectome might.

If you want a sample file it shouldn't be too hard to get one, but I believe there are various projects that offer them - conectome might.

+1 for the quality and safety of MRI

> they'd quickly replace ultrasound during pregnancy

Since scan time goes up with detail, and babies are really squirmy, it would be interesting to see how a quick-scan (low resolution) MRI compares with ultrasound quality.

But ultrasound destroys MRI in terms of portability, availability, and cost: just wheel in the cart, apply some goo, and there's the baby.

Perhaps more serious situations should use MRI more often instead of diagnostic ultrasound.

I expect that portability is actually a reason why often even visual ultrasound isn't used. Doppler scanning is often all that's needed to monitor heart rhythms

Squirmy babies shouldn't be too much of an issue. They do cardiac MRI all the time and hearts are hopefully beating during those ;)

Those are much easier because heartbeats are regular, breathing is regular, and you can image a bunch of breaths and heartbeats and synthesize a nice clean movie.

Motion in general is harder.

Better still, stop the breathing. Breath hold sequences or turning off the anaesthetic machine - both work better than free breathing.

Ah that makes a lot of sense. I guess it's pretty hard to schedule (f)MRI appointments around a fetus's sleep.

Random motion is far harder to control for. Irregular heart beats cause a lot of trouble, as do moving babies. So imagine foetal cardiac MRI. There are new methods using some pretty mind boggling maths (at least it is tonne) that deal with movement very well. But as with all clever MRI stuff, you can have nice, or you can have quick.

Less safe for an infant though since they have to be sedated in order to stay still.

On fetal MRIs:

"It also does not require intravenous contrast material (dye) or sedation."

http://www.utswmedicine.org/stories/articles/year-2016/fetal...

Fetuses are relatively easier than infants because they are effectively strapped into the womb and don't even know they're getting an MRI. They can obviously wiggle around some, but don't move as much.

Infants are harder:

http://www.medscape.com/viewarticle/499273_11

My daughter had to have a second, sedated MRI because she moved too much during her first one.

> There's no safety reason why you shouldn't do this.

When you have seen a few done wrong you disagree with this. They have an inherent potential for death and injury and incidents are frequent, unlike US. Then there is the whole implant safety and contrast safety side of things.

People won't be replacing pregnancy scans with MR any time soon, and the cost of the machine is not the primary reason for this. They scan too slowly and pregnant women aren't that keen on lying still for ages. It's sometimes even unsafe to encourage extended periods of lying supine. US is just too cheap and readily available.

Source: I do MR scans.

Young children need sedation so they will stay perfectly still; definitely not without risk.

My 1 yo son needs a scan every 3 months and the hospital had to figure a better way to scan him so as to avoid sedation. They finally settled on ultrasound: faster, risk free and precise enough once you know what you are looking for.

Reply on news.ycombinator.com