Building an MRI machine (for $100)

[e100]

Had an eye and brain MRI scan today in one of those million dollar superconducting magical bits of electronic gear. Despite being fitted with ear plugs and ear protectors it was like being in a loud electro rock concert - it sounded like the machine was trying to tear itself apart.
Afterwards I started reading about MRI on Wikipedia, and wow, talk about mind blown, no wonder people got Nobel prizes.

I don't have a million dollars, but does anyone know if $100 would be enough to build something that just shows the principle at work, or is this kind of thing just too hard on a tiny budget?

Mike
   

[LukeW]

This is not MRI (it's more like CT, with X-rays) but it's still a beautiful project.

http://www.tricorderproject.org/blog/dr-jansen-or-how-i-stopped-worrying-and-learned-to-love-the-barium/

[Artlav]

Depends on which principle you want to see.
NMR can probably be demonstrated fairly cheaply with a pair of big neodimium magnets and the pulse/receiving electronics.

To get an image is a whole another set of orders of magnitude of complexity.

Here is a nice approach to getting it working for cheap using coded sensing, but it haven't produced any results yet.
https://hackaday.io/project/5030-low-field-mri

Another approach can be to use easier to detect materials for special purpose scanners.
I.e. http://www.technologyreview.com/news/409887/a-fuller-picture-of-your-lungs/page/1/ - imaging of lungs with a DIY-grade cheap MRI machine.

[ale500]

Like Artlav said, you can demonstrate the principle very cheaply. You need a magnet. A magnet from a not so old harddisk are know to have about 1T.
Here you will find a math-physics-chemistry-heavy explanation: http://www.cis.rit.edu/htbooks/nmr/inside.htm

It looks like a quite fun project.

This book also provides very useful information: https://www.biophysics.org/Portals/1/PDFs/Education/james.pdf

But I'd like to (yet) find a book that describes how to build a continuous wave machine, or pulse-based one. As far as I found out the "pulse" is a very short pulse and not a burst of few cycles... but that is something I am not certain. 

For a 1H NMR with a 1 T magnet, you get 42.85 MHz and 60 MHz with 1.41 T . If you would use CW.

But the pulse approach is much better, you send your pulse to the coil and you digitize what you get in the other coil. After a FFT... you get you spectrum. A MRI is quite a few steps away 

Happy hacking !

[LukeW]

If you want real NMRI, your first step on that path would be the ability to show NMR effects in a test tube - essentially building your own NMR spectrometer, and going from there towards MRI.

http://kuchem.kyoto-u.ac.jp/bun/indiv/takezo/opencorenmr/

http://www.conspiracyoflight.com/NMR/NMR.html

(The latter website seems a bit nutty in places, but the NMR page is sensible enough.)

[AndreasF]

Like the others have said, "NMR" as in "nuclear magnetic resonance" should be possible to demonstrate with a relatively cheap setup. The really tricky bit is the "I" in MRI, as in "imaging", as that requires repeated pulses and acquisitions, a number of additional fairly precise (gradient) coils, and a bit of math (look up "k-space").

[T3sl4co1l]

Getting a sufficiently uniform field to really observe 1H line width is quite a challenge (I doubt a hard drive magnet will be anywhere near good enough), but it's nonetheless quite achievable.

Here's the classic citation (I have a PDF copy handy):
Bloembergen, Purcell, Pound; Relaxation Effects in Nuclear Magnetic Resonance Absorption, Phys. Rev., 73, no. 7 (1948)
The authors (and others) have related articles earlier and later. One (I don't happen to remember what, but I believe it was also in Phys. Rev. a few years before this) even shows the splitting effects of the different hydrogens in ethanol, founding the study of chemical NMR.

The apparatus is pretty primitive, though achieving field uniformity requires precision, adjustment, and a lot of poking around.  They used an impedance bridge approach, which can be nulled in both amplitude and phase, detecting the NMR signal as an imbalance in the "antenna" leg of the bridge.  Signal source was a radio signal generator (it will have to be stable in the Hz/minute range, so a good quality LC oscillator is needed, or even crystal controlled).  Signal amplitude was amplified and detected with a radio receiver, and read out on an oscilloscope.  Sweep was done magnetically, with a small variable DC bias coil driven by an amplifier and low frequency oscillator.  (Of course, frequency swept sources weren't so practical back in the day, so they didn't bother with that method.)

These days, usual process is to zap the nuclei with an RF burst (in-plane from the same coil, or out-of-plane from a perpendicular coil), and measure the FID (free induction decay) with an RF downconverter, DAQ and FFT.  The signal bandwidth is <100Hz, so a quite aggressive downconversion can be used (like going from FID at 30.000MHz down to audio frequency with a 29.998 MHz local oscillator and mixer).  Multiple channels (X-Y-Z coils) can be used to gather more information about precession and relaxation time.

Tim

[ale500]

The signal bandwidth is <100Hz

100 Hz means 10 MHz... to see aldehydes you need to look around 10 ppm...

What is actually brilliant and kind of helped in the old days is that the splittings due to H-H couplings lay in the ~ 1-10 Hz range, and thus could be seen with early machines right away. Nowadays, with modern ~500 MHz machines, you hardly see the couplings unless you zoom....

At least it is doable. I wanted for the last 10 years or so build a mass spectrometer. The problem are... the vacuum pumps. You need a turbo and they cost an arm and a leg .

[Kleinstein]

Just to demonstrate NMR one could also do zero Field NMR: that is measure the resonance in the earth's field. This is more like a sensitive absolute magnetometer. This needs a reasonable size coil (e.g. 500 g of copper), something like 0.5 L of water and the amplifier/ power source. The frequency is in the 1-10 kHz range.

[akis]

Maybe not $100, but a portable machine would make the inventor a very rich man. The reason is the MRI machines cost a fortune to buy, so they are only available in large hospitals, and even if the machines are 30 years old, they still charge an arm and a leg to run an MRI.

For example 30 years ago a doctor came down to the A&E floor carrying his portable (shoulder strapped) ultrasound device, complete with screen and probes and performed an ultrasound scan on the fly without moving the patient. Cannot remember if it was battery operated.

I understand probably 0.0001% of how the MRI machine works, and would really appreciate a simple ? explanation

[Alexei.Polkhanov]

Just to demonstrate NMR one could also do zero Field NMR: that is measure the resonance in the earth's field. This is more like a sensitive absolute magnetometer. This needs a reasonable size coil (e.g. 500 g of copper), something like 0.5 L of water and the amplifier/ power source. The frequency is in the 1-10 kHz range.

I think you are talking about Proton Magnetometer, which is very sensitive magnetometer. I have seen few of DIY magnetometers like that - fun part comes when you pair it up with GPS and map some large area. You can find things like ancient/old buildings...

One thing is - local university UBC had working version of MRI machine for doing Quality control of vine in bottles. This maybe another take on same approach:
http://livedigitally.com/wine-tech/ Guy was trying to turn his small MRI machine into business. I think it is awesome idea, not sure if it turned into working product.

[e100]

Found an MRI "how it's made" video on youtube
Skip to 3:17 for the construction of the gradient coil and 5:23 for one of the types of RF coil.

There's also a 40 second sound clip at
which in reality carries on for 20 minutes or more whilst you are expected to remain perfectly still. I was also supposed to keep my eyes still which is harder than it sounds when you have nothing to focus on and stressed by the noise and claustrophobia of being wedged inside a metal tube.

[T3sl4co1l]

Wow, they want $2 for that episode...

[Artlav]

Wow, they want $2 for that episode...

Huh?

[Alexei.Polkhanov]

Wow, they want $2 for that episode...

Because you are in US? Strange ...

[Kleinstein]

NMR can be used in different ways:
1) Measure the magnetic filed  (e.g. proton magnetometer)
    - this is relatively easy, it needs sensitive amplifiers and some DC power for excitation
2) Do some kind of chemical / physical analysis of a sample (e.g. the test of wine bottles)
    - this needs a very homogeneous, usually strong field. This might be a high quality permanent magnet arrangement.
      Excitation needs a sizable amount of HF power.
3) MRI - this is making pictures, using a magnetic field with well defined inhomogeneity.
    This needs a very strong magnet, control over the field and quite a lot of HF power.