Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

A low-cost and shielding-free ultra-low-field brain MRI scanner

nature.com

Post reply on HN
Print view View original

11–20 of 151 posts

Re: A low-cost and shielding-free ultra-low-field brain MRI scanner

#11

Post by rexreed »

So you're saying we have sufficient trust in the same sort of NN technology that confuses 8's and 0's in OCR text will be used to impute image data which might or might not exist? Sure, NN's are great at "filling in the gaps" and colorizing pictures based on what might be assumed, but when accuracy matters, does this approach truly work?

EDIT: I just want to point out that the original subject title of the post on HN was "A low-cost and shielding-free ultra-low-field brain MRI scanner Using AI" ... and the Using AI part of the post title was subsequently removed.

Re: A low-cost and shielding-free ultra-low-field brain MRI scanner

#12

Post by scratcheee »

a-dub wrote: post #5

woah. woah. woah. hold on a second here... are we comfortable enough with understanding all of the behavior of deep learning models to where we can confidently put them in the pipeline for diagnostic clinical imaging? i'm okay with using them for image analysis, but denoising and other image production tasks seems dangerous. how do you know what you're looking at is real as opposed to something that just looks convin…

Exactly my thoughts too. I'm fine with a simple noise-removal pass, but if the AI is context-aware, what's to stop it saying "hmm, this brain would look more like a normal brain if I remove these tumors". Obviously, they'll test for that, but that only handles common cases they concider, it's always going to be a risk for more unusual sceanrios, and the danger with altering the data is that anyone looking at the results wont have a way to tell how dubious that data is.

Reminds me of https://en.wikipedia.org/wiki/Xerox#Character_substitution_b... which was _so much_ worse than the equivilent OCR bug because it occured at the image level, where everyone expects errors to to produce noise, not contextly sensible and sharp _but wrong_ characters.

EDIT: based on other comments below, this is thankfully not the case, the AI just understands noise, it doesn't try to "fill in the blanks" based on how brains are supposed to look.

Re: A low-cost and shielding-free ultra-low-field brain MRI scanner

#13

Post by egocodedinsol »

“There are approximately seven scanners per million inhabitants and over 90% are concentrated in high-income countries. We describe an ultra-low-field brain MRI scanner that operates using a standard AC power outlet and is low cost to build.”

This is fantastic. What a sentence to get to write.

Re: A low-cost and shielding-free ultra-low-field brain MRI scanner

#14

Post by robertlagrant »

scratcheee wrote: post #12
a-dub wrote: post #5

woah. woah. woah. hold on a second here... are we comfortable enough with understanding all of the behavior of deep learning models to where we can confidently put them in the pipeline for diagnostic clinical imaging? i'm okay with using them for image analysis, but denoising and other image production tasks seems dangerous. how do you know what you're looking at is real as opposed to something that just looks convin…

Exactly my thoughts too. I'm fine with a simple noise-removal pass, but if the AI is context-aware, what's to stop it saying "hmm, this brain would look more like a normal brain if I remove these tumors". Obviously, they'll test for that, but that only handles common cases they concider, it's always going to be a risk for more unusual sceanrios, and the danger with altering the data is that anyone looking at the resu…

> a simple noise-removal pass

Even that is inventing data, no?

Re: A low-cost and shielding-free ultra-low-field brain MRI scanner

#15

Post by tshaddox »

scratcheee wrote: post #12
a-dub wrote: post #5

woah. woah. woah. hold on a second here... are we comfortable enough with understanding all of the behavior of deep learning models to where we can confidently put them in the pipeline for diagnostic clinical imaging? i'm okay with using them for image analysis, but denoising and other image production tasks seems dangerous. how do you know what you're looking at is real as opposed to something that just looks convin…

Exactly my thoughts too. I'm fine with a simple noise-removal pass, but if the AI is context-aware, what's to stop it saying "hmm, this brain would look more like a normal brain if I remove these tumors". Obviously, they'll test for that, but that only handles common cases they concider, it's always going to be a risk for more unusual sceanrios, and the danger with altering the data is that anyone looking at the resu…

Hopefully the system would be trained to accurately convey relevant medical information rather than to generate an image of a brain that looks normal.

Re: A low-cost and shielding-free ultra-low-field brain MRI scanner

#16

Post by iancmceachern »

a-dub wrote: post #5

woah. woah. woah. hold on a second here... are we comfortable enough with understanding all of the behavior of deep learning models to where we can confidently put them in the pipeline for diagnostic clinical imaging? i'm okay with using them for image analysis, but denoising and other image production tasks seems dangerous. how do you know what you're looking at is real as opposed to something that just looks convin…

Yes. This is past tense, other companies are already doing this, in the clinic.

Re: A low-cost and shielding-free ultra-low-field brain MRI scanner

#17

Post by xattt »

scratcheee wrote: post #12
a-dub wrote: post #5

woah. woah. woah. hold on a second here... are we comfortable enough with understanding all of the behavior of deep learning models to where we can confidently put them in the pipeline for diagnostic clinical imaging? i'm okay with using them for image analysis, but denoising and other image production tasks seems dangerous. how do you know what you're looking at is real as opposed to something that just looks convin…

Exactly my thoughts too. I'm fine with a simple noise-removal pass, but if the AI is context-aware, what's to stop it saying "hmm, this brain would look more like a normal brain if I remove these tumors". Obviously, they'll test for that, but that only handles common cases they concider, it's always going to be a risk for more unusual sceanrios, and the danger with altering the data is that anyone looking at the resu…

Therein lies the dilemma of this technology: would a scanner that might sometimes substitute information be better than no scanner at all?

Re: A low-cost and shielding-free ultra-low-field brain MRI scanner

#19

Post by ortusdux »

l33tman wrote: post #8
a-dub wrote: post #5

woah. woah. woah. hold on a second here... are we comfortable enough with understanding all of the behavior of deep learning models to where we can confidently put them in the pipeline for diagnostic clinical imaging? i'm okay with using them for image analysis, but denoising and other image production tasks seems dangerous. how do you know what you're looking at is real as opposed to something that just looks convin…

This project doesn't use AI to improve the image, they use it to estimate the EMI noise from the surroundings. So they're not "filling in the gaps" in the actual resulting 3D voxel volume with fantasy voxels (which I hope will never ever fly in a clinical setting). "To tackle the EMI signals from the external environments and internal low-cost electronics during scanning, we developed a deep learning driven EMI cance…

I'm glad that this is the approach that they are taking. There have been plenty of issues with fMRI false positives due to misconfigured software.

The most famous would probably be the IG Nobel winning study that detected brain activity in a store-bought salmon:

https://blogs.scientificamerican.com/scicurious-brain/ignobe...

https://www.discovermagazine.com/mind/fmri-gets-slap-in-the-...

Later studies called into question the results of between 10% and 40% of historic fMRI studies:

https://blogs.warwick.ac.uk/nichols/entry/bibliometrics_of_c...

https://www.pnas.org/content/113/28/7900

Re: A low-cost and shielding-free ultra-low-field brain MRI scanner

#20

Post by tshaddox »

rexreed wrote: post #11

So you're saying we have sufficient trust in the same sort of NN technology that confuses 8's and 0's in OCR text will be used to impute image data which might or might not exist? Sure, NN's are great at "filling in the gaps" and colorizing pictures based on what might be assumed, but when accuracy matters, does this approach truly work? EDIT: I just want to point out that the original subject title of the post on HN…

Do you have any data on the reliability of OCR systems used in production? I don’t have any such data, but given that the USPS was using OCR to sort mail over 50 years ago I would be surprised if these systems aren’t incredibly accurate.
Post reply on HN