Title: Data assimilation on a whole brain using Brain/MINDS database
Ken Nakae, Department of Systems Science, Graduate School of Informatics, Kyoto University
Data assimilation is a statistical technique for integrating a simulation approach with various types of observational data. A numerical weather perdition is an example of the data assimilation, in which the state of the simulation model on the earth is updated using the huge observation data by a satellite. Recently, neuroscientists have obtained a huge amount of the neural activity, connectivity, and genetic data. One of the big databases is by Brain/MINDS project, in which neuroscientists in RIKEN CBS obtain functional MRI, diffusion MRI, Electrocorticography (ECoG), virus tracer and in situ hybridization data of marmosets. We will develop a method of data assimilation for integrating the various kind of the data and understanding a complex dynamics and functions of the brain. In this talk, I introduce a prototype of the data assimilation for forecasting and interpolating neural activities on the whole brain using ECoG and diffusion MRI data, which is important for predicting a se izure. I also provide a method with Wilson-Cowan model to integrate functional MRI and diffusion MRI data with a collaboration study (Tsukada et. al., 2018). The result suggests a multimodality of an excitatory and an inhibitory balance of the marmoset’s brain. Finally, I discuss a future plan for developing the data assimilation and its potential in neuroscience.