Communications of the ACM
Computer Can Read Letters Directly From the Brain
A new mathematical model has allowed scientists to analyze MRI scans of the human brain to reconstruct actual thoughts more accurately than previously possible.
Credit: Mind Freaking Technologies
Radboud University Nijmegen scientists say they used a mathematical model to analyze magnetic resonance imaging (MRI) scans of the human brain and reconstruct thoughts more accurately than ever before.
Researchers in the Donders Institute for Brain, Cognition, and Behavior "taught" a model how small volumes of 2x2x2 mm from brain scans, or voxels, respond to individual pixels. By combining all the information about the pixels from the voxels, it was possible to reconstruct the image viewed by the subject.
The team used handwritten letters, and the result was a somewhat fuzzy speckle pattern. The scientists then gave the model prior knowledge, teaching it what handwritten letters look like, which greatly improved recognition. The images were then analyzed again. "The result was the actual letter, a true reconstruction," says lead researcher Marcel van Gerven.
"Our approach is similar to how we believe the brain itself combines prior knowledge with sensory information."
Researcher Sanne Schoenmakers says a more powerful MRI scanner will be employed for later research. "Due to the higher resolution of the scanner, we hope to be able to link the model to more detailed images," Schoenmakers notes.
From Radboud University Nijmegen
View Full Article
Abstracts Copyright © 2013 Information Inc., Bethesda, Maryland, USA
No entries found