Research

The brain is a tremendously complex system. In order to understand it we are going to need large amounts of data from many different kinds of measurements. We use data science methods to integrate the information provided by these measurements into a coherent picture. In particular, we develop statistical analysis techniques to decipher the role of networks of brain areas in complex behaviors and in brain disorders. We implement these techniques in robust, efficient, and openly-available computer software.

Human Connectomics

White matter connections between brain regions form a network that integrates information across the brain, comprising approximately 45% of the total cortical volume. Diffusion MRI (dMRI) is a technique that non-invasively measures properties of the white matter, assessing individual differences in the properties of the major tracts. We develop computational tools for analysis of dMRI data. Together with Jason Yeatman’s group at Stanford and Noah Simon in the Department of Biostatistics, we established DIRECT (Data Intensive Research in Connectomics), a research collaboration focused on use of large-scale datasets of human neuroimaging data to understand the brain and the complex relationships between the properties of brain networks, complex behavior and brain health.

Tools and publications:

Multidimensional analysis and detection of informative features in diffusion MRI measurements of human white matter
Adam Richie-Halford, Jason Yeatman, Noah Simon & Ariel Rokem
BiorXiv preprint

Combining citizen science and deep learning to amplify expertise in neuroimaging
Anisha Keshavan, Jason Yeatman & Ariel Rokem
Frontiers in Neuroinformatics, 13: 29

A browser-based tool for visualization and analysis of diffusion MRI data
Jason Yeatman, Adam Richie-Halford, Josh Smith, Anisha Keshavan & Ariel Rokem
Nature Communications: 9, Article number: 940

Vision science

We work on modeling and analysis of the biology of the early visual system. In some projects, we created phenomenological models of retinal processing, that simulate prosthetic vision. These models, implemented in software are useful for assessing the utility of these devices and as tools for designing new devices and stimulation protocols. In other projects, we analyze data about retinal health from large-scale databases. We are also interested to understand how the biology of networks in the early visual system evolves over time, and in response to disease.

Cloud-enabled data-driven discovery

Progress in understanding the brain depends on sophisticated analysis of massive neuroscience datasets. This requires the adoption of data science technologies that are emerging in industry, such as cloud computing. One of the objectives of our work is to reduce the barriers to wider adoption of these technologies. We are exploring multiple aspects of cloud computing. This includes development of software for deployment of computations to cloud systems, as well as work on web-based tools for data visualization and apps for citizen science. Together with Beth Buffalo and Adrienne Fairhall in the Department of Physiology and Biophysics, and in collaboration with the Jupyter team, we established a cloud-based platform for interactive computing in multi-electrode recordings from human and non-human primate brain.

Tools and publications:

Cloudknot: A Python library to run your existing code on AWS Batch
Adam Richie-Halford & Ariel Rokem
Proceedings of the 17th Python in Science Conference (2018): 8 - 14

A browser-based tool for visualization and analysis of diffusion MRI data
Jason Yeatman, Adam Richie-Halford, Josh Smith, Anisha Keshavan & Ariel Rokem
Nature Communications: 9, Article number: 940

Tools and publications:

Data science

As we apply tools from statistical learning to problems in neuroscience, we also end up developing general purpose open-source statistical computing tools. Our lab is also involved in a variety of ways in data science education and training.

Hack weeks as a model for data science education and collaboration.
Daniela Huppenkothen, Anthony Arendt, David W. Hogg, Karthik Ram, Jacob T. VanderPlas, & Ariel Rokem
PNAS, 115: 8872-8877