“Open Source Biomedical Imaging” - this legit?

[ivaylo]

https://www.crowdsupply.com/mindseye-biomedical/spectra
If legit can it be used at this resolution for something practical which can’t be done by other means? I see the apple and potato example, but it’s a bit underwhelming...

[CatalinaWOW]

Their kit has 32 electrodes, used in pairs, so only 16 spatial samples.  Do all the diagnostics you wish with a four by four image of your innards.  There are tricks they can play to get hyper resolution, and they sell upgrade kits to add electrodes,  but even after all that it is very low resolution as is demonstrated by their sample images.

[Domagoj T]

Looks like it has potential to maybe be capable of imaging tumors once they grow to the size of a small watermellon.
The comparison between lung scans done by this device and a comparable CATSCAN clearly show why the CATSCAN is prefered, despite the radiation.
They need to crank up the resolution by a few orders of magnitude if this is ever going to take off.

[harnon]

Cool idea, although I agree resolution would be an issue. Also are "the masses" really the best people to interpret biomedical images? I'd be surprised if somebody like Phillips or St Jude hasn't patented that absolute crap out of a tech like EIT for this application.

[jean123]

It is indeed legit. It's a technology also outside of the patent period. The spatial resolution is proportional to the electrode spacing and geometry(if you wish to scale up or down use different electrode arrangements).
https://en.wikipedia.org/wiki/Electrical_impedance_tomography
Also here are some established larger companies doing it in a clinical setting. Drager ( https://www.draeger.com/en_uk/Hospital/Products/Ventilation-and-Respiratory-Monitoring/ICU-Ventilation-and-Respiratory-Monitoring/PulmoVista-500) and Swisstom.

There is also a hackaday project showing its evolution over history: https://hackaday.io/project/159737-spectra-open-biomedical-imaging

It's cool because it's a cheap inexpensive way to see inside any conductive object, just modelling the body as a grid of resistors. CATSCANs use about 256 point to reconstruct an image, and this kit is using 32 - although 32 electrodes leads to 896 different projection combinations.

If you want to see what the first MRI looked like have a look at this page: https://two-views.com/mri-imaging/history.html  MRI has gotten a lot better since those initial images, and perhaps EIT(electrical impedance tomography) can do. It has the advantage that it can be a low cost, safe and portable device.

[SiliconWizard]

Indeed. Also, AFAIK, although not exactly an imaging technique, what is called "minute ventilation" is usually an impedance-based measurement in most pacemakers. This is related even though it requires a lot fewer electrodes and a lot less computation.

Now obviously with the above project, spatial resolution will be pretty poor.
As a side note, just be very careful with safety with this kind of devices if you're going to build and test homemade prototypes on living beings.

[rhb]

The technology is valid.  The abilities of the developers is dubious at best.  Compressive sampling can provide a 5-10x improvement in resolution per sample.

Here's a recent paper by one of the the mathematicians responsible for developing compressive sensing:

https://statistics.stanford.edu/sites/g/files/sbiybj6031/f/2017-09.pdf

Emmanuel Candes is his cohort in crime.

[jmw]

I've seen a live demo of it at a Hackaday event in SF. The resolution is limited as noted but the kit seemed well designed.