Google AI and Harvard researchers unveil intricate images of the human brain through 3D mapping technology

A collaborative effort between Google and Harvard University researchers has led to the unveiling of highly detailed images of the human brain, showcasing its intricate neural networks through advanced AI and 3D mapping techniques. This groundbreaking research, a decade in development, involves Google Research’s Connectomics team and Harvard’s Jeff Lichtman among others. The project utilized a significant dataset derived from a brain sample donated by a woman undergoing epilepsy surgery. Over 5,000 tissue slices were imaged at high resolution by Jeff Lichtman’s team using a multibeam scanning electron microscope, a process taking 326 days.

Google’s team contributed by integrating these images into a cohesive 3D model, highlighting each cell’s structure. This synergy of brain imaging and AI-driven analysis allowed for the reconstruction of nearly all cellular connections within a portion of the brain tissue, roughly half the size of a grain of rice. The resulting images offer unprecedented insights into the brain’s operational intricacies, potentially aiding in the understanding of neurological disorders.

The project’s neural network imagery, possibly the largest dataset of human brain structure to date, required over a million gigabytes (1.4 petabytes) of data to illustrate just a small section of the brain. The researchers focused on a part of the cortex known as the anterior temporal lobe, applying color coding to neurons based on size and type to enhance visibility and identification for further study.

Analyzing a minuscule tissue sample, about one cubic millimeter in volume, the team identified approximately 50,000 cells and 150 million synapses, highlighting the dense network of connections fundamental to the nervous system. Notably, some neuron pairs exhibited up to 50 synaptic connections, a phenomenon the researchers are keen to explore further.

Among the intriguing findings were ‘axon whorls’ – axons that, instead of extending outward, looped or coiled. These uncommon structures add another layer of complexity to our understanding of neural connectivity and signal transmission within the brain.

This collaboration between Google and Harvard University has not only pushed the boundaries of neuroscience research but also underscored the potential of combining AI and 3D mapping in unraveling the mysteries of the human brain.