Michigan Neuroimaging Initiative

About the YBRC Brainbox

The Yottabyte (YBRC) project is a cloud-based service housed at the Univeirsity of Michigan and providing interactive computing resources to researchers. Virtual machines (VMs) are created and accessed as ‘workstations in the cloud’.

The YBRC Brainbox project combines the Brainbox system with the YBRC infrastructure to create a flexible, low-cost system for neuroimaging analysis.

Each VM comes with a reasonable amount of local storage, and large network storage volumes can be provided by ARC-TS or ITS.

YBRC Brainbox configuration

A typical YBRC Brainbox VM consists of four processors, 12 GB of memory, and 1 TB of local disk space. This is a good setup for a small lab where one to three people will typically be working on a machine.

If more processing or more memory is needed, machines can be shut down and reconfigured with additional processors and/or memory and simply restarted.

Connecting to a YBRC Brainbox is done using SSH and VNC, which provides a secure connection and a graphical interface. Sessions can be detached, and upon reconnection, the desktop appears as it did when detached.

We also plan to make available configuration scripts that would enable someone to configure a generic installation of the supported operating system as a Brainbox, outside of the YBRC. This would be appropriate for use on a workstation in a lab. Only the free and open source software would be included to insure that licensing terms are fully complied with.

YBRC Brainbox software

We intend to provide a spectrum of neuroimaging software on each YBRC Brainbox. As of the time of writing, the list of software includes Matlab, SPM 12, SPM 8, the Psychiatry Methods Core software distribution, FSL, AFNI, the Anaconda2 Python distribution, R, R Studio, ANTs, and ITK-SNAP Convert3d. Additional software will be added as the project progresses.

We are interested in exploring the full range of neuroimaging software that is available. Many tools seem to be used primarily because they have been used. It would be useful to all if we could discuss how to properly evaluate new tools, or new versions of existing tools, to see whether they should be recommended (or warned against). What should be done when one software package produces results that differ from those of another? How robust are the methods and their various implementations? Introspective topics like these are likely to become more relevant and important in the coming years.

If you are interested in getting more information about the project, getting a demonstration, or signing up for a YBRC Brainbox, please contact brainbox-help@umich.edu.