A team from Radboud University in the Netherlands has developed a system that utilizes functional magnetic resonance imaging (fMRI), imaging software, and a specialized algorithm to identify through a person’s brain scan exactly what they are looking at.
Functional MRI scans, which are typically used to measure changes in brain activity, are used in this study to zoom in on smaller (2 x 2 x 2 mm), more specific regions of the brain in the occipital lobe called voxels, according to The Daily Mail.
Scientists had participants look at a series of letters—B, R, A, I, N, and S— while wired up to an fMRI scanner. By recording the specific changes that occurred in each person’s brain after each letter was shown, the scientists were able to show how the brain visualized the different shapes. The images captured of these changes in the brain were then run through an algorithm that was designed to work in a similar way to how brains build images of objects from sensory information.
An image of what the person is looking at is reconstructed when the algorithm converts the voxels and their relevant changes into image pixels. As of now, the model is only designed to compare letters, but it could be used for other imagery, such as a person’s face, and could also become more accurate, Marcus Van Gerven, co-author of the study, Linear reconstruction of perceived images from human brain activity, told The Daily Mail.
He also claims that the algorithm could be used to reconstruct any image, and that his team is working to build more advanced machines that can build images from 15,000 voxels. (The “letter” experiment used information obtained from 1,200 voxels).
Share your vision-related news by contacting James Carroll, Senior Web Editor, Vision Systems Design
To receive news like this in your inbox, click here.