[Introduction to New Wisdom]Just today, people were shocked by this 1 cubic millimeter nanoscale map of the human cerebral cortex. Google’s ten-year neuroscience achievement, the Human Brain Map, has also appeared on Science. Among them, scientists have discovered cells that have never been discovered before, and new connection patterns.
The human cerebral cortex can be modeled with nanometer resolution!
Advertisement
The connectomics team at Google Research has been established for ten years. As a commemoration, just today, the team released this 1.4PB connectome map of the human brain.
It was in this image that Google scientists discovered characteristic cells that had never been seen before.
The figure contains 57k cells and 150M synapses.
Advertisement
1 cubic millimeter of brain, mapped in stunning detail. The impact this picture brings to people is really shocking.
This 3D map covers a volume of about one cubic millimeter, one millionth the size of the entire brain, contains about 57,000 cells and 150 million synapses, and is a massive 1.4 petabytes of data
For 10 years, the Google Research connectomics team has been committed to using high-throughput methods to study neural network architecture in the brain and improve our understanding of brain structure and function.
This commemorative article can be said to be the culmination of the research results of the past ten years.
The article has been published in Science and reported by Nature.
Paper address:https://www.science.org/doi/10.1126/science.adk4858
What is the significance of research on human brain reconstruction?
Google scientists believe that if we continue to study brain connections, we may one day understand how our memories are formed and even find the causes of neurological diseases, autism, and Alzheimer's disease.
Netizens said: “These images are so shocking, it is like observing complex structures in outer space. Maybe we will not be able to understand everything in our own brains in our eternal life.”
Some people also said: “The human brain has 100 billion neurons and 300 trillion synapses, but GPT-4 only has 2 trillion parameters, so we still have a lot of room.”
Six brain neural maps with amazing results
The six brain maps shown next are all drawn with the help of Google AI. With this, we finally unveil the intricacies of the brain.
Visualizing through six layers of cerebral cortex
First, Harvard researchers collected thousands of extremely thin cross-sectional images from donated brain samples.
This small piece of healthy brain tissue was removed during surgery on an epilepsy patient to allow doctors to reach the area that needed surgery.
Subsequently, Google used the development of advanced AI tools to build an interactive 3D model of this brain tissue.
As shown below, this 3D model highlights the high complexity of the human brain.
Just this small sample (about 3 millimeters long and accounting for one millionth of the human brain's capacity) requires more than 1 million GB of data, or 1.4PB.
This is the highest resolution and largest data set on the structure of the human brain ever produced.
The sample came from a part of the cerebral cortex (gray matter) called the anterior temporal lobe (see image). The cerebral cortex has six layers, with neurons colored according to their size and type. In the magnified view of all neurons, neurons in each layer are clearly visible.The surface of the brain is at the top edge of the image
A dense “map”
One cubic millimeter of tissue sample contains approximately 50,000 cells and approximately 150 million synapses.
Some pairs of neurons have the amazing property of being so tightly interconnected that they have as many as 50 synapses connected to each other.
The image below shows a close-up of an excitatory neuron, colored by size, largest in red and smallest in blue. The core diameter of these cells is approximately 15-30 microns.
mirror dance
During the reconstruction process, the researchers also discovered a peculiar phenomenon. Cell groups often appear in a “mirror symmetry” manner.
Just like the picture at the beginning, this pair of cells seems to be dancing.
This layer contains so-called “triangular neurons,” which have one basal dendrite that is much larger than the other dendrites. 77% of the triangular neurons can be divided into two main categories: those with large basal dendrites tilted to one side of the sample, and those with mirror-symmetrical angles tilted to the other side.
Statistical analysis showed that neurons with the same tilt type were more likely to be located next to each other. This statistical correlation suggests that there may be some unknown underlying function at work.
swimming in synapses
Neurons in the brain are closely connected.
As shown below, this neuron (white) has more than 5,000 axons (blue) from other neurons, transmitting signals.
And there are at least an equal number of synapses, where signals pass from the axon to the receiving neuron.
Synapses are shown in green in the figure.
A strange discovery: axonal spirals
A peculiar discovery of this study is an “axonal spiral”, which is the blue part in the picture below.
Axons (blue) are the filament-like parts of nerve cells that carry signals out of the cell.
These ring-like stacks of axons are very rare in the sample, and in some cases they are located on the surface of another cell (yellow).
As for the function of this blue “axon spiral”, it is still unknown.
serious networking
The white part in the picture below is a single neuron.
It receives signals that determine whether the neuron is firing.
And this diagram shows all the axons it can be told to fire (green) and all the axons it can be told not to fire (blue).
Imagine how many such neurons there are in the entire brain. This is a huge amount of information!
Establishing a “brain map” at the cellular level
Although the function of most of the human body's vital organs is not much different from that of other animals, the special features of the human brain set us apart from other creatures on Earth.
The human brain is composed of billions of cells interconnected to form a neural network. It may be the most complex computing machine in existence. Its capabilities exceed many artificial computing systems that consume astonishing power, but its power consumption is only about 12W, and About the same as an incandescent light bulb.
Currently, our understanding of the human brain stops at which areas are responsible for which functions. If you want to further explore how it works, such as how memories are formed and the mechanisms of neurological diseases, you need to go down to the cellular level.
This is the research content of the emerging field of “connectomics”, which aims to understand and accurately reproduce the connections between brain cells and establish a “neuron map” of the brain.
The connectomics imaging method in this study used nanoscale resolution to reconstruct quadrillion voxel-level segments in the cerebral cortex, including 1,600 neurons, 32,000 glial cells, 8,000 vascular cells, 150 million synapses, but the actual brain tissue involved is only one cubic millimeter, equivalent to half a grain of rice.
To perform reconstruction work, we first need to collect image data of real samples, that is, the brain tissue samples of epilepsy patients mentioned above.
Using this sample, Professor Lichtman's team in the Department of Molecular and Cellular Biology at Harvard University produced more than 5,000 slices with a thickness of about 30 nanometers, and used a device called a “multi-beam scanning electron microscope” to collect high-resolution images. , the image acquisition work alone took 326 days.
The study used brain tissue samples thought to be located in the left anterior temporal lobe
On this basis, the team aligned and stitched the image data, reconstructed the three-dimensional structure of each cell, including axons and dendrites, identified synaptic connections between cells and classified the cells.
Using a small piece of human brain tissue, the researchers constructed a three-dimensional image of nearly all the neurons and their connections. The upper picture shows excitatory neurons and the lower picture shows inhibitory neurons.The tissue samples used were approximately 3 mm wide, with neuronal cell bodies ranging from 15 to 30 μm in diameter.
New discoveries in neuroscience
These reconstructions of the human brain are revealing never-before-seen structures that could change our understanding of how the brain works.
For example, research has discovered a rare but very powerful synaptic connection in which there may be more than 50 individual synaptic connections between a pair of neurons.
96.5% of the connections between axons and target cells contained only one synapse, but 0.092% contained four or more synapses. The study discovered the pattern of these connections, which, combined with statistical analysis, shows these strong connections are no accident.
Further study of these connections may reveal their functions in the brain, such as as a mechanism for rapid neural responses or a way of encoding important memories.
In extremely rare cases, a single axon (blue) forms multiple synaptic connections (yellow) with a target neuron (green). The purpose of these tight connections is unknown.
Since the brain tissue samples came from epilepsy patients, although the researchers did not observe obvious signs of pathology under a light microscope, they cannot rule out that these special structures are related to the patient's disease or the drugs he is taking. Perhaps more samples can be analyzed. Find out why.
The above findings may be just the tip of the iceberg. The research team stated that this data set is very large and complex, and it is believed that there will be more new brain structures and characteristics to be discovered in the future.
AI empowers brain science
Due to the rise of AI and the development of various software tools, connectomics is becoming more and more powerful.
Before the advent of AI tools, the first connectome, published in 1986, consisted of just 302 neurons in the nematode C. elegans and took researchers 16 years because they needed micrographs of cross-sections of all the nematodes. Cells were manually stained.
When the Google Connectomics team was established ten years ago, one of their visions was to use the cutting-edge results of AI to process huge data sets in the field of biology, so that they could develop from a nematode with 302 neurons to a complex system with tens of billions of cells. biological tissue.
The development of connectomics since the 1970s
With the support of AI, researchers no longer need to manually colorize 1.4PB of electron microscope data. They have developed an RNN model called “flood-filling” that can automatically segment electron microscope pictures and reconstruct nerve cells, and has relatively high accuracy. High accuracy, no need for a lot of manual proofreading work.
This article was published in Nature Methods in 2018.
Paper address:https://www.nature.com/articles/s41592-018-0049-4
On this basis, the team also developed the automatic recognition algorithm SegCLR, which is used to identify and classify cells in different parts of the brain neural network.
In order to store and manage massive multi-dimensional data sets, the Google team also launched TensorStore, an open source software library based on C++ and Python, which has been widely used. The project has received 1.3k stars on GitHub.
project address:https://github.com/google/tensorstore
References:
https://www.nature.com/articles/d41586-024-01387-9#ref-CR1
https://www.science.org/doi/10.1126/science.adk4858
https://research.google/blog/ten-years-of-neuroscience-at-google-yields-maps-of-human-brain/
https://sites.research.google/neural-mapping/
https://blog.google/technology/research/google-ai-research-new-images-human-brain/