Harvard and Google Collaborate to Map Human Brain at Subcellular Level

In a groundbreaking achievement, biologists from Harvard University, in collaboration with Google, have successfully mapped a cubic millimeter of human cerebral cortex at the subcellular level. This innovative project, which took a decade to complete, involved creating an interactive map of brain tissue that includes approximately 57,000 cells and 150 million synapses. The initial images were derived from a small sample of a woman's brain, and the researchers utilized electron microscopy to enhance visibility by employing heavy metals that reflect electrons. The project saw Google’s machine-learning team, led by Viren Jain, play a crucial role in aligning the 2D images into 3D reconstructions. By applying advanced algorithms for automatic segmentation, the researchers were able to differentiate various cell types, color-coding them for clarity—smaller cells in blue and larger ones in red. This method facilitated the identification of the brain's six cortical layers and white matter, although challenges remain in proofreading the automatic segmentation to ensure accuracy. Significantly, the research has revealed previously overlooked phenomena, such as a single neuron forming over 50 synapses, which could be vital for understanding cortical processing. The brain map has been made publicly accessible, heralding new opportunities in neuroscience research, particularly in understanding and treating mental health conditions like schizophrenia. Looking ahead, the researchers aim to explore additional human brain samples to further their studies and believe that insights gained from the biological brain could also enhance the development of deep-learning AI systems.