Engineers specialized in artificial intelligence developed the most detailed 3D brain map from 225 million images
A team of specialist Connectomics engineers from Google and the Lichtman laboratory of Harvard they managed to create the 3d brain map most complete in the world.
To create this map called H01 Dataset 225 million images were used. According to the researchers, it can be seen in real time and could help to understand how complicated the human brain is.
The brain map is a colorful 1.4 petabyte reconstruction, with nearly 4,000 incoming axons connecting to a single neuron.
Although the H01 Dataset It is the most complete, just a small part of our brain is shown.
In the design the cerebral cortex is represented so you can see how sophisticated cognitive functions such as thinking are processed: memory, language and perception.
What is the most complete 3D brain in the world like?
According to the research document, the images are of a cubic millimeter of brain tissue, with millions of neuron fragments, 130 million synapses surveyed and more.
In order to develop this elaborate brain map, the scientists divided the sample into “5,300 fragments 30 nanometers thick. Which gave the computer the 225 million separate 2D images to analyze and finally merge into one of the most elaborate 3D maps of the human brain. “
According to the experts, they have information on “250,000 nerve cells, tens of thousands of reconstructed neurons, millions of neuron fragments, 130 million annotated synapses and other structures (axons, dendrites, etc.)”.
Despite all the work, Google He noted that the brain map barely covers one “millionth of the volume of a complete human brain.
In this sense, it was specified that there is still much work to be done to discover the functions that the human brain can develop.
“Although the image reflects only one millionth of the size of the map of a human brain. The volume of memory required to store this data reaches 1.4 petabytes, almost a million gigabytes “
Why see the brain map the platform must be accessed Neuroglancer. Although it is very complete, it is still necessary to study the architecture of the brain network.
“Neuroglancer is a WebGL-based viewer for volumetric data, capable of displaying arbitrary cross-section views of volumetric data. As well as 3D meshes and models based on line segments “