In a sample of what he is capable of achieving The fusion of AI and Neurosciencea team of scientists from the University of California in San Francisco (UCSF) and the Allen Institute, designed an AI model that has created one of the most detailed maps of the mouse brain to date.
This new map includes previously unexplored brain subregions, which opens new ways for neuroscientific knowledge. The findings have been published in Nature Communications. This brain cartography offers an unprecedented level of detail and expand our understanding of the brain By allowing researchers to link specific functions, behaviors and diseases with smaller and more precise cell regions, which provides a road map for new hypotheses and experiments on the role that these areas play.
“It’s like moving from a map that only shows continents and countries to one that shows states and cities,” explains Bosiljka Tasic, director of molecular genetics of the Allen Institute and co -author of the study. This new and detailed brain plot, based solely on data, and not on the annotation of human experts, reveals previously unexplored subregions of the mouse brain. And, thanks to decades of neuroscience, new regions correspond to specialized brain functions still to be discovered.
The basis of this advance is Celltransformer, a powerful AI model that can automatically identify important brain subregions from massive spatial transcriptomic data. This technique reveals the location of certain types of brain cells in the brainbut cannot show brain regions based on its composition.
Now, thanks to Celltransformer, Scientists can define cerebral regions and subdivisions based on calculations of “shared cellular neighborhoods”similar to drawing the limits of a city according to the types of buildings that compose it.
“Our model is based on the same powerful technology as the tools of AI as chatgpt. Both are based on a framework of ‘transformers’ that stands out for its ability to understand the context -adds Abbasi -Sl We use Celltransformer to analyze the relationship between nearby cells in space. Learn to predict the molecular characteristics of a cell based on its local neighborhood, which allows you to create a detailed map of the general tissue organization. ”
This model successfully replicates known regions of the brain, such as the hippocampus; But, more importantly, You can also discover previously not cataloged sub -regions and more detail in little -known brain regionsas the reticular core of the midbrain, which plays a complex role in the beginning and release of the movement. What distinguishes this brain map from others?
The new brain map represents brain regions compared to cell types. Unlike previous brain maps, Celltransformer’s is completely based on data, which means that their limits are defined by cell and molecular data, rather than by human interpretation. With 1300 regions and subregions, it also becomes one of the brain maps based in data mtos granular and complex jamtos performed on animals.