In the field of neuroscience, a revolutionary breakthrough is quietly taking place. How complex is the structure of the human brain? How powerful is it? Recently, Google and Harvard University brain science researchers teamed up to successfully model 1 cubic millimeter of brain tissue at the nanoscale, and actually stored a staggering 1400TB of data.
The amount of data we usually come into contact with rarely reaches the level of TB, because 1TB is equal to 1024GB of capacity, which is basically the hard disk capacity of today's ordinary computers, so 1400TB is equal to 1.4336 million GB, and this is only a cubic millimeter of brain tissue modeling data stock.
The volume of one cubic millimeter of brain tissue is also the size of a millet grain, but scientists have found more than 57,000 cells, 150 million synapses, 230 millimeters of tiny blood vessels, etc., and countless fine structures, which makes people marvel at the complexity of the brain tissue structure.
The reconstruction, called H01, is the first to demonstrate synaptic connectivity networks in the human brain at nanoscale resolution. At the heart of this study is a sample of temporal cortical tissue from a 45-year-old female with epilepsy, about 1 cubic millimeter in size. Through a series of delicate processes, including rapid fixation, staining, and resin embedding, the samples were cut into thousands of ultrathin sections, each with a thickness of only 33.9 nanometers. The slices were then imaged using a multibeam scanning electron microscope to obtain raw 2D image data with a total size of about 1.4 PB.
With this ultra-high-resolution imaging technique, scientists were able to see details that had never been seen before, including subtle connections between neurons and complex structures inside cells. Through the analysis of these structures, the researchers identified the major cell type composition of this brain region and discovered some previously unknown neuronal morphology and connections.
With such a large amount of data available, the researchers used computational tools to stitch together and align the 2D images and reconstruct the 3D voxel data. Next, they used a machine learning algorithm to segment the morphology of neurons into voxels, and by manually correcting the segmentation errors, they finally constructed the three-dimensional morphology of all cells, synapses, blood vessels and other structures in the 1 cubic millimeter brain tissue.
For example, they found that the vast majority of axons form only one synapse with their target cell, but a few axons can form as many as 50 or more synapses, creating particularly strong connections to the target cell. This "strong connection" of multiple synapses is prevalent in both excitatory and inhibitory axons, and its number is significantly higher than what would be expected when synapses are randomly formed. This discovery provides new clues to our understanding of the transmission of information between neurons and the brain's ability to process information.
In addition, the model also reveals the vascular system in the brain, which has important implications for studying the brain's energy supply and metabolic processes. By looking at the relative position of blood vessels and neurons, we can better understand the effects of blood flow and oxygen supply in the brain on neuronal activity.
However, the challenges of this research are no less enormous. First, the size of the human brain is much larger than this 1 cubic millimeter sample, and modeling the entire brain requires enormous amounts of data and computational resources. It is estimated that modeling the entire human brain would generate up to 1.76 zettabytes of data, which is far beyond the storage capacity of today's most advanced supercomputers.
Second, the number of neurons in the human brain is hundreds of billions and the number of synapses is as high as quadrillion, which makes it extremely difficult to perform accurate neuronal morphological segmentation and connection analysis in the model.
Nevertheless, this research still provides us with valuable insights and opportunities. This achievement not only refreshes our understanding of brain complexity, but also provides a valuable data base for future neuroscience research. It marks the deepening of human research on the human brain, and also provides a new perspective for us to understand how the brain works.
Through continuous technological innovation and algorithm optimization, we are expected to achieve nanoscale modeling and in-depth analysis of the entire brain in the future. This will open up new possibilities and avenues for us to understand how the brain works and prevent and treat neurological diseases.
At the same time, this research also demonstrates the great potential and value of artificial intelligence in neuroscience research, and provides us with new tools and ideas to explore the unknown world of the brain.
Source: "Global Network" reported on May 13 that "Google and Harvard join forces to draw the most complete and largest map of the human brain"