Non-Invasive Deep Brain Stimulation

Mess with your head by wearing a hat!

We’ve known since the time of Galvani that you can use electric fields to stimulate neurons and make bodies do things.

Stimulate a nerve in a frog’s leg, and you can make it twitch. Stimulate the right part of a human’s motor cortex, and you can make them raise their arm; stimulate the right part of their visual cortex, and you can make them see a bright light.

Ultimately, the problem with drugs that act on the brain is that they’re carried by the blood, and they go everywhere. You can’t selectively “turn on” one brain region and not another. We do crude things like messing with “serotonin” or “dopamine”, molecules that have tons of different functions, all over the brain and body.

In principle, electromagnetic stimulation to just the right neurons could be a hugely more flexible and precise paradigm for therapeutics.

The best-researched examples of brain stimulation being useful for actual diseases tend to be deep brain stimulation, in which an electrode is surgically placed at a region deep in the middle of the brain.

For instance, stimulation of the globus pallidus is an established treatment for refractory Parkinson’s disease. In Parkinson’s, the dopamine-producing neurons in the basal ganglia die off, leaving profound problems with the ability to start and stop motion and speech. You can supplement dopamine with L-Dopa, but that only works for a while, because you’ll start producing even less dopamine to compensate. As I understand it, the premise of brain stimulation is “let’s compensate for the loss of most of your dopaminergic neurons by making the remaining ones more active.”

In animals, electrical stimulation of deep brain structures can have really striking effects.

By stimulating different parts of the hypothalamus, you can increase or decrease appetite and thirst, increase desire to exercise, cause fight and flight reactions and sexual behavior, induce puberty and restore male fertility in old age, alter body temperature, and manipulate the circadian rhythm.

Could hypothalamic stimulation be used to reverse obesity? There hasn’t been much experimentation in humans, mainly because the risks of brain surgery make it not worthwhile for otherwise-healthy people. But it’s mechanistically very plausible.

Other deep brain structures have been investigated as electrical stimulation targets. The nucleus accumbens, which is involved in motivation and reward, can be stimulated to reduce addiction and OCD symptoms, as well as a few case studies where it reduced obesity. Stimulating the thalamus may help traumatic brain injury patients regain consciousness.

The trouble with deep brain stimulation, of course, is that it requires brain surgery. It would be easier to do human experiments if there were a safe, non-invasive way to stimulate deep brain structures.

Non-invasively electrically stimulating the surface of the brain isn’t too hard — you can do it with electrodes, what’s called transcranial direct current stimulation, or tDCS. But these signals don’t penetrate very deep. You’re basically limited to the cortex.

Transcranial direct current stimulation at safe doses (<2 mA) doesn’t directly cause neurons to fire. Rather, it creates an electric field of about 0.4 mV/mm which alters the polarization of neurons slightly, making them either more or less excitable depending on the polarity of the electric field. Notice that the electric field attenuates rapidly in space.

Electrode stimulation, by contrast, usually uses alternating current, and directly induces neuron spiking above around 100 Hz. And the electrode is ideally right next to the neurons you want to stimulate. Transcranial alternating current only goes as far as the cerebral cortex.

Back in 2017, Ed Boyden’s lab created a workaround, called temporally interfering electric fields.

The key is, neurons are not activated by high-frequency electrical signals (>1kHz) but they are activated by low-frequency signals. So if you have two very high-frequency electric fields outside the skull, at slightly different frequencies, say 1000 Hz and 1040 kHz, then they will have interference patterns forming a wave with a frequency of 40 Hz. This wave will be largest in amplitude where the fields interfere constructively, which may be at a point distant from the electrodes on the head — deep within the brain.

They tried it in a mouse. Interference between two electrodes at 2000 Hz and 2010 Hz made neurons fire at 10 Hz, just as well as if you’d put a 10 Hz electrode right at those neurons. Kilohertz stimulation alone did not cause neuron firing. By moving around the locations of the two skull electrodes, they could focus the electrical stimulation to the hippocampus deep within the brain; hippocampal neurons fired at roughly 10 Hz. Using c-fos expression as a marker for neural activity, they found that hippocampal activity increased but cortical activity didn’t. In other words, the interference effectively targeted stimulation to the hippocampus.

The stimulation was given at 125 uA, well under the safety threshold for transcranial stimulation, and it produced no elevated rates of neuron death, change in synapse density, or activation of microglia. It looks safe, at least after a single 20-minute session.

Now, what has been done since then?

And, how likely do we think this is to work in humans?

Human heads are bigger than mouse heads and their skulls are thicker. It may be necessary to increase the amplitude of the electrode signals in order to get through all that brain tissue — will this still be safe?

A computational study using MRI-derived models of three young men’s brains estimated that amplitudes of 0.8-1.25 mA were optimal for directing stimulation to the hippocampus. These are comparable to amplitudes used in tACS, and thus reasonably safe. Unfortunately, however, the peak amplitude at 10 Hz in the hippocampus was barely higher than the amplitude at 10 Hz in the cortex. In other words, the stimulation was not very spatially precise. It may be possible to target brain regions more precisely by using more electrodes.

Another human brain model simulation study predicted that with four electrodes at 1 mA each, regions of interest in the pallidum, hippocampus, and motor cortex could reach field strengths of 0.24-0.5 V/m, high enough to induce firing.

The bottom line is, our best guess is that we won’t need dangerously high currents to get effective deep brain stimulation, but there’s a lot of computational and electrode density work yet to be done to get the stimulation site to be precise.

Other experimental applications seem promising — in a rat model of opioid overdose, temporal interference stimulation could restart the diaphragm and cause the rats to start breathing again. Neither a single electrode at high frequency (5000 Hz) or low frequency (1 Hz) had the same effect; the temporal interference was necessary to penetrate to neurons deep in the spinal cord.

Also apparently some Chinese researchers have straight-up tried temporal interference stimulation on human subjects. 2 mA electrodes, at 2000 Hz and 2070 Hz, focused on the motor cortex, reduced reaction time on a motor task. Now, admittedly, it’s not deep brain stimulation, but at least it suggests that the thing works at all.

So, currently, I’m pretty optimistic. Precision would have to improve in order to make temporal interference stimulation competitive with traditional invasive deep brain stimulation, but that’s such a heavily computational problem that I expect it to improve over time.

And maybe wearable deep-brain-stimulating therapies will become a reality.