Neuralink From Elon Musk Is Literally Mind-Blowing
It is possible to do an internal neural communication? It is possible to stimuli and detects neural activities with an electrode in the brain? These questions are expected to be answered in this article.
First, this article was written in sections, to facilitate the reader experience. Some sections have compacted information, other more specific details in neuroscience or engineering. It is totally up to the reader to go thru all the sections or not. For a complete understanding of the technology and applications mentioned in the title, the author truly recommends reading the whole structure.
Brain Activity
This section presents the main idea of the neurological system, introducing the action potential theory, neuron structures, etc.
There are 86 billion neuron cells in the typical human brain. Neurons send and receive information. Although neurons come in many different types, they generally have three parts: a dendrite that receives a signal, a cell body called soma which computes the signal, and an axon which sends a signal out.
Neurons are connected through synapses. The neurons of the brain connect to each other to send and receive signals through axon-dendrite connections called synapses.
Neurons communicate through electric signals. Action potentials cause synapses to release neurotransmitters. These small molecules bind to receptors on dendrites, opening channels that cause current to flow across the neuron’s membrane. When a neuron receives the ‘right’ combination of spatiotemporal synaptic input, it initiates an action potential.
It’s possible to record electrical signals in the brain. To get this, electrodes should be placed near neurons in order to detect action potentials. Recording from many neurons allows to decode the information represented by those cells. There are neurons in the brain that carry information about everything we see, feel, touch, or think.
Neural Engineering
This section presents the notion of the importance of future technology applied in medicine, more precisely in the engineering of the brain.
When speaking about technology in healthcare, specially orientated to hardware systems, it talks about biomedical engineering. Instrumentation, controlling, automation, sensing and etc, are the main fields to create new technology in medicine, f.e., robotics (this include prosthesis, bionic members, electromyographs), machines for image diagnosis (ultrasound scan, MRI), and human sensors (pulse oximeter, electrodes).
Neural engineering is a branch of biomedical engineering and by its complexity, is totally acceptable that, everything from electrical engineering, neuroscience, biochemistry and others subjects, is all compacted into it. Getting started on electrical engineering point, since simulations to manufacture the first prototype, is required hard mathematical modeling, conditioning circuit, signal processing, power supply effectiveness, a study on interference based on electromagnetic dynamics, simulations and etc. After that, it must be integrated with the other areas to be acceptable. It is not that simple.
Neural engineers are interested in understanding, interfacing with and manipulating the nervous system. (…) One benefit of understanding this communication is to provide new ways to interface between neural tissue and manmade technologies. This is known as brain-machine interfacing. IEEE(EMB)
But let’s suppose that the idea is feasible. Imagine an implanted device inside a head that could detect electrical signals from human body, especially on brain, neurological signals, brainstem connections, and the spinal system. Imagine that it can process visual, audition, motor activating sensor signals on this device. After that, picture this kind of tiny electrode applying an ideal and controlled stimuli pulse for a compromised neural network, the exact network that is responsible for the movement of a body member, or touch sense, visual network, and turning the paraplegic patient able to do a step, or blind into a watcher, or people with communication problems into speakers. Even more, send all those signals, responses, wireless to a computer, from the skin.
I can say that this is totally getting out of the “paper” and going to some “heads”. Yes, and it is not a brand new Black Mirror episode.
Elon Musk is beginning a new era of neural engineering. Not impossible for who’s already CEO of, a smart space rocket industry (SpaceX), and a robust, powerful, and effective vehicles industry (Tesla). He revealed Neuralink’s plans for brain-reading “threads” and a robot to insert them in an online Conference (August 2020).
Neuralink develops and approaches the idea behind the brain-machine interface (BMI) or similar to the brain-computer interface (BCI) either. Instead of EEG (Electroencephalogram), a non-invasive method for neural monitoring, that results in blurred signals (mostly because of detecting difficulties), Neuralink shows the invasive model more effective, handling with secure and body compatibility.
Actually, Elon is the Neuralink founder and majority owner of the company, but the CEO is Jared Birchall.
Neuralink Overview
This section shows the overview of the structures of Neuralink technology.
Compared to the link of the last year (2019), the new device got some improvements. One to highlight is, the recharging evolution.
- Electrodes — has the same size scale as neighboring neurons and as flexible as possible; there are dozens of electrodes in a “thread” (a signal conductor and insulator of 43 milliliters long and totally flexible); microfabrication of the threads results from thin film metals and polymers resistant to corrosion from fluid in the tissue; it must have sufficient surface area to allow stimulation;
- Chips — the Link of Neuralink needs to convert the small electrical signals (microvolts) recorded by each electrode into real-time neural information; it must have high-performance signal amplifiers and digitizers;
- Packaging —Link needs to be protected from the fluid and salts that bathe surrounding tissue on the brain; it’s very hard when that enclosure must be constructed from biocompatible materials; multiple components with a process that builds them as a single component is presented;
- Neurosurgery — the threads are too thin to be manipulated by hand and too flexible to go into the brain on their own (imagine trying to sew a button with thread but no needle); it needs to safely insert them with precision and efficiency; the solution is based on a new kind of surgical robot.
Threads, which are smaller than human hair and are 4 to 6 micrometers in diameter, can transfer huge amounts of data, and 32 threads contain 1024 electrodes per array.
The Neuralink chip has the inductive charge method. To get the Neuralink, the robot puts in the electrodes and does the automated surgery in less than one hour, making the patient possible to leave the hospital on the same day and without general anesthesia.
The first thing that Link will do is detect neuron activity, by using signal processing to control an outside computer. As the users think about moving their arms or hands, it could decode those intentions, which would be sent over Bluetooth to the user’s computer interface.
The Neuralink app would allow to control an iOS device, keyboard, and mouse directly with the activity of the brain, just by thinking about it.
We plan to help people with severe spinal cord injury by giving them the ability to control computers and mobile devices directly with their brains. Neuralink
Experiment
This section shows three important experiments with three pigs (very similar structures to the human brain) done at the last Neuralink conference presented by Elon Musk.
The first experiment shows that the first pig, Dorothy, remains healthy and happy, after an implant removing procedure as a result of the factual reversibility of the chip.
The second experiment shows the spikes and sounds beeps of neural connection with the chip. When the other pig, Gertrude, snuffles around and touches something with her nose. The blue wave at the figure below, shows the accumulative spikes of activation from electrodes connections with neurons, in white dots, on time. Gertrude had been with the link for over two months, with health.
The last one shows the comparison of the measured motor activity and the predicted with computer vision of the shoulder, elbow, carpal, and trotter joints, in order, of the pig in a treadmill. Yes, a pig on a treadmill!
The experiment results show the good functionality of the implanted system.
Neuralink Detailed
This section shows detailed information from Neuralink chip operation modeling, in terms of hardware and software functionality. It will detail starting from the “macro” to the micro way point of view.
The first Neuralink project is detection. As the neuron fires, the spike (or action potentials) connects to one of the 1024 electrodes channels. The pulse needs to flow into a passband filter (instead of a time domain, there is a frequency domain, so there is a bandwidth frequency range of the neural spike that needs to remain) to exclude noise signals.
For a better manipulating, this filtered signal needs to be amplified digitized by an analog to digital converter (ADC). After processing it on an internal microprocessor, the signal must be converted in digital to analog (DAC) and, finally, amplified be wireless transmitted, as Neuralink engineers confirm, Bluetooth low energy radiofrequency (2.4 GHz).
The following image illustrates a simplified flowchart (from the author’s point of view) for the signal and the small idea of the BCI initial process (neural detection and stimulation process).
Some additional blocks could be included in the chart. For example, in the process before the transmission switch, could be mixed an oscillator, with the output amplified signal, for a better signal modulation in a specific frequency. This procedure allows a better telecommunication signal processing, in terms of power and signal-to-noise factor. Or even a conditioning circuitry block for power supply. For simplicity, are not included such specifications on the flowchart.
All the processes in the flowchart in terms of sensing, amplification, conversion, and processing, are totally integrated on the N1 chip of the Neuralink. N1 is a totally customized ASIC (Application Specific Integrated Circuits) for the brain's specifications, assuming that there is nothing on the market that provides such features, as neural amplifiers f.e.
The duration of an action potential is one millisecond. N1 digitizes it at 20 kHz, so the hole signal is divided into 20 pieces to process.
The ADC divides the magnitude of each piece in 1024 levels, so N1 has 10 bits of resolution (as the digitized conversion is only about ‘1’ and ‘0’, total bits is 2¹⁰ = 1024). The detection process duration is, 900 nanoseconds (0.0000009s), and this is determined by the internal clock of the N1. Considering neural speed communication, everything on Link happens faster than the brain!
The entire Neuralink measures 23 x 8 mm. Going inside of N1 SOC (System-On-Chip), which has 5 x 4 mm, it’s possible to find 1024 channels for all electrodes. Each channel has a noise correction factor of 7.2 micro Volts. For each thread channel, there is an analog pixel with 6.6 micro Watts of power consummation.
Over time, Neuralink improved three version revisions of the analog pixel, in size, power consumption while maintaining its performance. The last version on the right, (see figure below), is five times smaller than the known state of the art. Each pixel is dedicated to each electrode, as published in academic literature.
Let’s zoom in more, going from the analog pixel channel thru the thread with the electrodes. A hair has a 100 microns of diameter, the thread called the linear edge, has the electrodes, whereas each one has 5 microns! Neuralink made over 20 designs of the threads, and the goal was to increase the number of the electrode at each channel, without significantly increase the width of these threads at the base (see figure below).
However, as the number of electrodes increases, these raw digital signals become too much information to upload with low power devices, and, in this case, N1 essentially saturates in signal processing, and may cause wrong measurements and overheating f.e.
The surgery robot must precisely do the incision and implant exactly on specific spots that do not harm the bloodstream.
Speaking on software, there isn’t one programming language describing all algorithms for the Link and the surgery robot. To have an idea, at the low levels of chip programming, in terms of the registers and peripherals structures, it uses Verilog, but it also uses C, C++, Python, Java…
Final Considerations
This section presents some considerations about Neuralink's future visions and the author’s opinion.
- Control or insulate the brain electric field of the entire implant ought to be accomplished, even detecting spikes or stimulating them;
- Decrease material resistance for cleaner detection, as the electrode detects the action potential spikes. This procedure will optimize the filter and amplifier;
- Decreasing the thickness of the electrode is a huge gain, but it results in a reduction in the cross-section of a conductor, directly influencing the flow of electrical current. So the main objective is to optimize these correlated functions;
- Elon Musk revealed that, Link including surgery procedure, it will cost a few thousand dollars and will decrease it over time;
- One thing to prevent is the internal and external interference, such that sealing the Link package must be as better as it can possible, mainly the electronic circuit insulation;
- In the conference was discussed that in the future it will possible to make a backup and replay the memories. In additional, probably will be able to export communication that is hard to expose, like ideas, engineering models, art, music, using f.e. augmented reality;
- Put an end on diseases (control chemical, enzyme productions), cure anxiety, depression, autism, and etc.
- The Neuralink will process the external cortex, in special motor, sensing, vision, auditory, and language areas. But when talking about emotions, consciousness, psychological behavior, at the moment, will be limited to detect and study. Those are located in profound regions, like the hypothalamus. To reach these regions isn’t reasonable just extend a thread. Other scientists present at the conference related that brain movements, physical impacts, and the whole structure must be studied. But it is possible.
Connect with LinkedIn!
