Cool work and proof of concept, and very excited to see where this goes. However, I do think there is enough exaggeration and missing information here that it warrants some critical appraisal. What's really missing is a comparison and validation with any existing medical imaging tech. Whole brain, contrast-free neurovascular imaging is essentially solved with MRI, why not run a scan and compare? Ultrasound is of course portable and less expensive, but MRIs are actually widely available in most cities at reasonable cost for medical workflows, and low-field brain MRI is addressing the portability and cost issues to some extent. I guess they are pitching this as a wearable "telepathy" device, which I think appropriately differentiates their product, but of course, this wording also invokes a framing that "you won't / don't need to know how it works," which invites skepticism and a higher bar for validation in my view.
"MRIs are actually widely available in most cities at reasonable cost" - I live in one of those first-world countries, and our citizens regularly wait many months if not over a year to get a single MRI scan. Yes, it's not just an issue of the MRI but the entire medical system, but the point still stands. Were there machines that were one or more orders of magnitude cheaper and simpler to run - I think we would see a marked increase in availability.
I agree on your ground-truth desire, and I would hope they've done a lot of that to validate what we see here.
I completely agree that's an issue, although more of an economic / public health policy issue than a technical one. There are low field MRI systems, such as the one made by Hyperfine that are, like you say, an order of magnitude cheaper and simpler to run. We should have these everywhere, IMO
MRIs are fundamentally expensive. Yes we can bring the price down a bit, and we can set more money aside for them, but they’ll always be limited by their price.
Even if this technique is much worse (I can certainly believe it is) the price might allow uses that would never be practical with MRI even with the best financial support. For example, ultrasound might be viable for use in GPs or small medical facilities which could never dream of justifying an MRI machine.
Yes totally, and ultrasound already does wonders in that regard. It's a good strategy to focus on the specific use cases that match the strengths of the tech. I think MRI will be useful in validating and mapping out those cases.
> Wow, in the USA, we can get MRI's the same day or at worst case, week.
You can. And the cost is higher than almost anywhere on earth.
You can get them quickly in most places with a publically funded healthcare system, it’s just that a priority patient is very very sick and you never want to be that person.
Yes, that is a economic & public health policy problem that really needs to be solved. We can look to Japan as example of what's possible, they have invested in nearly twice the number of scanners per capita of Canada, and they can get same-day MRIs for $50, roughly speaking.
The super-resolution trick as they’ve done it is highly reliant on the sparseness of the bubbles. If you imagine a point or a very sparse set of points at low resolution, you can fit for the locations of those points even though you don’t see them clearly. This is a common technique in radio astronomy and (I assume although I don’t have personal knowledge) astrometry, and compressed sensing was an extremely hot field a while back.
But RBCs are weird squishy things, and they fill the bloodstream quite densely, and ChatGPT estimates that they’re spaced about 20µm apart and that, when confined to a capillary, they’re about 7µm long. (And that sounds at least plausibly correct to me.)
So, even ignoring the much worse scattering properties of RBCs, they not nearly as sparse. You mostly lose a whole dimension of sparseness and up trying to resolve the entire capillary. Which seems possible but much harder. Unfortunately, brain capillaries are about 40µm apart, so the result might be a mess.
The article did not say what wavelength they’re using or what their native (wavelength/2) resolution is.
Showing us a technique that is entirely reliant on sparseness and then saying they hope to employ it on something that isn’t sparse at all (blood cells) does feel misleading.
I’m filing this in the category of technologies I wish could be true, but for which no plausible path to overcoming the obvious limitations has been provided.
From the bubble center plot, I'm guessing that the bubbles are separated on average about a few mm apart? Taking the other comment's guess at face value, you're going from about 2 mm to 20 um, so 2 orders of magnitude. Air (technically SF6 in the article) and water (RBC is close enough) have acoustic impedances that differ by 3.5 orders of magnitude.
My assumptions here are *extremely generous*, i.e. favorable to the "oh, we'll just make it work with natural contrast", and even then, they can't hand wave 5-6 orders of magnitude of improvement. Furthermore, because of the use of super resolution, I'm guessing there's some exponential factor in there, i.e. double the density of bubbles/tracking points past some critical limit, then you need 8x the data to reconstruct things.
You have about 5L of blood, so that’s three orders of magnitude more volume than the contrast, and RBCs are 10x-50x as concentrated as the microbubbles in the syringe, so about 4 orders of magnitude concentration difference.
It’s basically changing this from a 0D problem to a 1D problem.
> From the bubble center plot, I'm guessing that the bubbles are separated on average about a few mm apart?
The page is vague so I can't tell. I think the images they're showing are actually a composite of many bubbles tracked through the vasculature.
They say this:
> As bubbles flow through the vasculature, we accumulate millions of these positions and stack them into a single image with detail finer than the wavelength.
And the rendering showing the bubble centers they're tracking only shows a few small points moving at a time.
I think that the amazing animation they produced at the top is actually a composite of many different trackings, not an actual representation of what they capture in real-time.
I’m a complete layman to this field, but what the article did say was they’re hopeful that AI/ML can help develop a model that can pull out information such as the scattering caused by RBCs (which is present in the large volume of data gathered by the probe but is too weak to be used for manual techniques) and turn that into meaningful visuals. That’s gonna require a ton of data and that is exactly what they are trying to gather now with what they have built so far.
> The bubbles themselves are pockets of sulfur hexafluoride encapsulated in lipid shells.
The high resolution images were generated by injecting sparse bubbles of this contrast agent. How sparse are they? Is the image we see a stacked set of many bubbles over time composited together?
Their aspirations at the end of doing this without the bubbles are great, but there’s a big “now draw the rest of the owl” energy around that leap. The first technique relies entirely on the bubbles, but they provide no explanation for how they think this could be achievable without the bubbles other than vaguely saying that technology is advancing.
Per Wikipedia, it "is a colorless, odorless, non-flammable, and non-toxic gas."
When used as a contrast agent for ultrasound, it "has been used to examine the vascularity of tumours" -- which would be similar to its use in the OP. Then
"[i]t remains visible in the blood for 3 to 8 minutes, and is exhaled by the lungs."
So -- not collected and excreted by the liver, as I at first thought.
Contrast-enhanced ultrasound (CEUS) with microbubbles has been around clinically for 20+ years. There are many contrast-agent manufacturers, e.g. Bracco (SonoVue/Lumason) or GE Healthcare (Optison). Safety-wise it's probably better than CT iodinated contrast or gadolinium MRI agents, and it's pretty well-established at this point.
Is Aleph an established entity with a track record that should lead us to trust this at face value? I couldn’t find any info on them and the site seems new
> None of us were ultrasound scientists before this. We worked backwards from a desire for brain interfaces and taught ourselves physics, ultrasound, electromagnetism.
Not to say it’s not interesting or neat, but the Silicon Valley approach to solving medical issues doesn’t have a good track record, let’s put it that way
The damage done by Elizabeth Holmes at Theranos goes way further than just that company. There is a lot of distrust now in anything tech that touches on medical devices. Some of it is for good measure, some of it will prevent really cool stuff from happening.
Exactly. The concept itself; a machine that can do a bunch of tests from a whole lot less blood would be amazing but anyone who wants to do this now is automatically "oh so like Theranos" and then not gonna give you money to do this PhD or post doc and do you figure out a way to do this? You can't raise money because everyone's gonna be thinking of Theranos.
The Midjourney scanners don’t do the same thing that this is using. See how blurry the first image on the page is? That’s what you get from ultrasound through bone like the skull.
They used a trick to inject sparse bubbles into the patient and let them flow through the brain, then looked for the perturbations caused by those sparse bubbles.
The Midjourney scanners aren’t injecting this bubble contrast agent into everyone’s veins.
Meta is also going at it [0], which inevitably makes me ponder some orwellian questions for the near future:
If I bring my pet mouse to the cinema and my friend scans the movie back using his apple ifmri does the DRM still holds or will the mouses be DRM locked?
Will my iris suffice for booting my computer or would I need to press accept all brainwave cookies?
Can I email my local Flock representative to install a new Brain Pole in my neighborhood? I saw a bunch of dark thoughted young males around and my amazon think camera says the probability of missing packages increased.
Certainly the thing of sci-fi nightmares, but not practical.
All of these imaging techniques are very involved. Ultrasound requires direct contact and this technique only works with a long IV infusion of bubbles. fMRI isn’t going to be a portable device that you can point at something for many reasons.
The connection to what you’re thinking is more sci-fi than reality. This technique could theoretically see some changes in blood flow to different regions, but what would that mean? Is the patient having anxiety, or are they just nervous about the IV injecting bubbles into them to travel to their brain and the machine attached to their head?
The current state of the world, where we have insane and ubiquitous surveillance tech but our packages are nevertheless being constantly stolen (with the thieves "caught" under said surveillance but with no one bothering to enforce it), is certainly an interesting one.
The surveillance is not about protecting you or your property; I'd argue it never has been. It's about protecting those in power and entrenching the power they wield.
I thought the whole "we can guess what you're thinking from an MRI" thing was BS, along the lines: take a small set of photos, image people's brains as they are looking at these pictures, to map to some low-dimensional vector of "brain activity". Then show them some of these (in sample!) pictures, measure the vector of activity and predict back what they were looking at.
Happy to be corrected. But if that's right then this... does the BS thing in a potentially less intrusive way?
Every few years one of these ultrasound companies comes around and promising to revolutionize medical imaging and nothing ever comes of it. Anyone remember https://www.openwater.health? The same ideas are in a perpetual state of being reinvented and part of me thinks its just a hustle for the MIT Media Lab/Stanford Imaging grads to give them something to do.
The tell is "super resolution", "brain computer interface" and "mixed modality" -- adding some contrast agent here, or maybe an IR light source.
It turns out the nyquist limit, diffraction and physics are real things.
The same thing has been said about robotics, AI, space travel, etc. etc.
I'm not saying this is the way, and I have significant questions of understanding thought based on reading brain activity, but I wouldn't put down the entire ultrasound field.
How about just getting it more established in orthopedic practices so patients aren't required to 1. See ortho for MRI referral 2. schedule mri at imaging facility 3. PAY $750 - $3000 for an MRI 4. Wait to get back into ortho.
I really don't understand why a fetus' heart can be examined for defects, but you can't use it in the office to tell me if my labrum is torn?
Why in the heck was this comment downvoted? Because that was exactly my thought reading the article: the mind machine interface stuff is weird (and fMRI blood flow is never going to achieve a lot it is a blunt tool which this is related to).
But high resolution imaging of blood flow? That's a pretty great medical diagnostic tool if you can make it more available and cheaper.
Additionally, as an ambulance chaser who looks at medical bills all day, people don't realize how much of the zombie medical debt out there is from scummy ERs (HCA etc) doing 2 or 3 pointless MRIs at $5k a pop.
This is ridiculously cool, but I have a ton of questions.
> The bubbles themselves are pockets of sulfur hexafluoride encapsulated in lipid shells. They're an FDA-approved contrast agent,
Combined with ultrasound, could these be causing damage of any kind to the vasculature?
> A few years ago, a paper came out that blew our minds. The idea was that you can decode what someone is looking at just from their brain activity.
How realistically close can this get to reading thoughts, visuals, etc.?
Do we have a path to imaging people's visual cortex? Their inner lives, dialogues, memories? (Scary thought - this could be used as an interrogation tool without consent. "Did you kill Bob?" could be a simple brain scan.)
Can it be done in real time in a feedback loop and perhaps be used as an advanced reinforcement learning system?