Decoding brain signals

Wayne Gillam

Almost everyone knows someone who has experienced a form of brain-related injury or disorder. There are thousands of people every year who undergo neurosurgical procedures, and many face a lifetime of impairment with limited treatment options.

The CSNE and its industry affiliate, Microsoft, are working together to advance neurological treatments by tackling one of the greatest challenges in neuroscience today—how to accurately interpret brain signals. Using a multi-pronged approach, the CSNE and Microsoft are developing and evaluating cloud-based analysis of electrocorticographic (ECoG) signals from the surface of the brain, using machine learning to interpret data. Recent collaborations include:

  • An online competition sponsored by Microsoft, Decoding Brain Signals, which builds on the dissertation work of Dr. Kai J. Miller, a University of Washington graduate who was supervised by the CSNE director, Dr. Rajesh Rao, and neurosurgeon, Dr. Jeff Ojemann. The competition uses Dr. Miller's dissertation work and data from his patients with epilepsy as a foundation for participants' projects. Decoding Brain Signals asks competition participants to build machine-learning models using the Microsoft Cortana Intelligence Suite, software that will decode perceptions from human subjects' ECoG signals. The learning model contestants create should predict whether the human subject is seeing the image of a house or of a face, based on ECoG signals collected from the brain surfaces of four patients with seizures. The contest provides $5,000 in cash and prizes for the winning entries.
  • A proposed Microsoft-sponsored hackathon, to be held at the CSNE. The hackathon will be designed to encourage the CSNE students to enter and compete in the Decoding Brain Signals competition.
  • Planning a brainstorming session between Joseph Sirosh, Corporate Vice President of the Data Group at Microsoft, his team, and the CSNE neurosurgeons and researchers to explore how cloud data/machine learning services could be of use to hospitals and doctors.

Brains connected with sensors to the cloud, and machine learning could revolutionize neuroscience and neurological treatments for those who have experienced injuries due to epilepsy, spinal cord damage or stroke. View an overview of this effort by watching a keynote presentation by Joseph Sirosh, Corporate Vice President of the Data Group at Microsoft. Learn more about the Microsoft contest, Decoding Brain Signals, by visiting the contest website and/or learn more about the CSNE's partnership with Microsoft by contacting Scott Ransom, PhD, the CSNE's Industry Liaison Officer. Microsoft's Decoding Brain Signals contest ends June 30, 2016.