Is the LimeSDR sensitive enough to pickup RF from brain activity?

So if an EEG is measuring macroscopic electrical activity taking place in the brain, then given the relationship of the electromagnetic spectrum, the brain should presumably also be producing RF. Would the LimeSDR be sensitive enough to receive these RF signals, if for instance you were inside of a faraday cage?

I want to get like an Intel Skylake server with a C236 chipset, which has 10x USB 3.0 ports and then connect 10 LimeSDRs around the brain with some type of faraday cage surrounding it. Each LimeSDR has 6 RX U.FL connectors and 2x2 MIMO, so you should be able to wire up like a 60 antenna array around the brain and then do an FFT frequency scan in parallel and such. Then I want to feed all these signals as sensory input into a deep learning neural network comprised of about 30k CUDA cores.

This should be capable of spatial filtering using computational signal processing to function as a three-dimensional EEG of the brain, to precisely locate the source and propagation of epileptic activity within the brain’s neural networks.

No -

I won’t argue with the fact that the signals are very weak, but I think it’s foolish to think that the brain only operates on delta, theta, alpha, beta, and gamma brainwave activity. The visible spectrum is from about 430 THz to 770 THz, and the eye’s are electromagnetic spectrum quantum mechanical sensors that feed the brain’s neural networks with up to about 187 terabytes per second worth of data, according to the Shannon–Hartley channel bandwidth theorem. Other animals can actually see in the infrared range as well, and still further magnetoreception has already been demonstrated to exist within other animals too. So I don’t think they are even close to the right ball park. The thing is though, you don’t really need to know the exact frequencies, because if you know anything about square waves, FFT harmonics will get you the input signals that you want, and once you have those signals the deep learning computer neural network which you are feeding the RF sensor data into will find the cause and effect relationship between the input and output. All you may need are some visual, auditory, and motor test patterns to calibrate the brain’s neural networks to correlate with the computer’s neural networks. Then you can do FFT spectrum analysis using beamforming and a three-dimensional computational model for mapping out the neural networks.

Then if that actually works, the next logical step would be transmit beamforming for low-power transcranial stimulation, and then if that works perhaps the next step could be non-invasive ablation of the areas that trigger the epileptic activity in treatment resistant patients.

I’m afraid not. LimeSDR-USB is indeed 2x2 MIMO, but this means 10x of them would give you 20x channels and not 60x. There are more U.FL ports than Tx/Rx channels because each channel has multiple ports that are optimised for a particular frequency range, which are then selected as part of configuration.

RF is generated by varying a electrical charge moving in a conductor. The human Brain does contain neurons which do conduct electrons in synaptic connections at various macroscopic neural oscillations (frequencies), that have historically been given a number of names:

  • (𝛿) delta waves (0 Hz to 4 Hz)
  • (θ) theta waves (4 to 8 Hz)
  • (α) alpha waves (8 to 12 Hz)
  • (β) beta waves (12 to 30 Hz)
  • low gamma waves (30 to 70 Hz)
  • high gamma waves (70 to 150 Hz)
  • (γ) gamma waves (30 to 200 Hz)
  • high frequency waves (200 to 500 Hz)

So the frequencies of those oscillations are in all likely-hood going to be the only ones that a brain could naturally emit detectable levels of electromagnet emissions. The eyes are probably another good example of the kind of signalling rate that happens within the brain, because we typically can not differentiate information from spinning objects that are rotating faster than 60 times per second (that would correspond to a frequency of about about 60Hz). Just because your eyes are transferring insane amounts of data/information, does not mean that the clock rate or transfer rate has to be very high at all with a ridiculously large number of parallel channels (each human optic nerve contains between 770,000 and 1.7 million nerve fibers and connect to as few as 5 photoreceptor cells in an eye).

For the RF world, bands of frequencies have also historically been allocated names:
ELF (Extremely Low Frequency)
Frequency range: 3 Hz to 30 Hz
Wavelength range: 100,000 to 10,000 km (62137 to 6214 miles)

SLF (Super Low Frequency)
Frequency range: 30 Hz to 300 Hz
Wavelength range: 10,000 to 1,000 km (6214 to 621 miles)

ULF (Ultra low frequency)
Frequency range: 300 Hz to 3000 Hz
Wavelength range: 1,000 to 100 km (621 miles to 62 miles)

A typical adult human brain needs about 20 watts *(refer to note1 below) of energy to function. There are about 100 billion neurons in a human brain, and each neuron has on average about 7,000 synaptic connections to other neurons, so at most each individual synaptic connection might be radiating nearly 30 femto-watts of energy ( -105 dBm). In reality that is probably an optimistic maximum value, but at least it is some kind of a starting point, using quick back of an envelope maths.

Then if you factor in Johnson-Nyquist noise, the bandwidth of your system would need to be less than 6MHz if the front end amplifier is operating at room temperature or even less bandwidth if your antenna has a lower efficiency (P.S. It will have a low efficiency at these frequencies). Like you would NOT be using a quarter wavelength dipole antenna (e.g. at 3 Hz, a quarter wavelength antenna would be about 25,000 km ; 15,534 miles in length - might make for an interesting visual prop in a science-fiction film, but it is not practical), you would use a much much much smaller antenna, which would be much much much lower efficiency) . At those kind of frequencies you would not use a traditional E-Plane antenna (electric field - think dipole antenna), you would use a H-Plane antenna (magnetic field - think large coil of wire).

Another possibility is that maybe RF is emitted by the 100 million neurons at a much higher signal strength than the synaptic connections. Then if you ignore all attenuation by the fat (about sixty percent of the mass of a brain is typically fat), water (blood) in your head and the walls of your skull, (like I did previously) that would be maximum of about ~200 pico-watts (-67 dBm) radiating away from each neuron before attenuation.

Basically if it was something that interested me I would be looking towards the more exotic RF equipment that is used for the reception of Schumann resonances. Basically low resistance coils of wire wrapped around mu-metal (exceptionally high magnetic permeability) cores for the antenna(s); multiple stage low bandwidth, ultra low noise amplifiers to bring the signal level up high enough so that they can be detected, and then I’d probably be looking towards extremely high the audio equipment (or at least the chips that are used for that) to digitise the signals. Usually for detecting Schumann resonances equipment would be setup in remote locations about 10 km (~6 miles) from human infrastructure, but that would not be necessary for your use case since you actually want to attenuate all man-made and natural RF sources.

There was a good presentation on The Royal Institution’s youtube channel earlier this year called "The Spike: How Your Brain Uses Electrical Impulses to Communicate - with Mark Humphries" (well in reality it was the author enticing people to buy their book “The Spike: An Epic Journey Through the Brain in 2.1 Seconds”, but it was still an interesting talk). To measure the spikes they used electrodes inserted as close as you can get to clusters of neurons inside the brains of living animals. The two take aways from it, for me, are that the brain is mostly doing nothing (while using energy) all the time and while doing nothing there are also paths in the brain where there is always activity all the time, almost like a clock signal, or a train of pulses, or propagating waves. It reminded me a bit of train carriages trundling along the tracks, irregardless if the carriages are empty or have passengers.

note1 (from above):
Men need around 10,500kJ (2,500kcal) of food a day. Women need around 8,400kJ (2,000kcal) of food a day.

2% of the mass of a human body is the brain, and about 20% of the daily energy intake is used by the brain.
10,500kJ/24/60/60 ~= 122 watts, and about 20% of that is about 24 watts.
8,400kJ/24/60/60 ~= 97 watts and about 20% of that is about 19 watts.