AI Unveils Your Visual Thoughts
Radboud University researchers have built an AI system to regenerate highly accurate images visualized by humans by decoding and recreating visual experiences directly through the monitoring of brain activity.
According to Umut Güçlü, a researcher at Radboud University in the Netherlands, these types of developments are innovative ones and described them as, “the closest, most accurate reconstructions.”
It is not the first time researchers have tried to use AI to understand what people see by looking at their brain scans.
Previous similar researchers conducted by Güçlü’s team used specific MRI machines known as Functional Magnetic Resonance Imaging (fMRI) to record the brain activity of three people who were looking at three different photos.
In another study, the brain activity of a macaque monkey was also recorded by linking tiny devices to its brain while looking at AI-generated images. According to Thirza Dado, a member of Güçlü’s team, the monkey was used for other research purposes and not reconstruction of images, stressing ethical considerations.
Functional Magnetic Resonance Imaging (fMRI) works by recording blood flow and oxygen levels in brain areas taking part in visual processing. Then, the data will be analyzed by AI, which in turn learns how to interpret brain signals and identify key regions where the images seen by individuals must be recreated.
The team analyzed again the data provided by these studies using an advanced AI system, which can determine brain regions capable of regenerating accurate images.
Güçlü explained, “Basically, the AI is learning when interpreting the brain signals where it should direct its attention,” adding, it mirrors the way the brain records details related to the environment.
The outcomes were impressive especially with direct recording to the brain, to the extent that some recreated photos were very similar to the AI-generated one seen by the macaque, generated by the StyleGAN-XL image-generating. In contrast, Dado emphasized that it was much easier to reconstruct AI-generated photos than real ones, and this is due to the AI’s capabilities and known aspects of the image generation process.
StyleGAN-XL is an advanced AI model specialized in the generation of high-quality images based on simple texts descriptions, by improving enhanced versions of images and making them more realistic.
Brain Implants Restoring Vision
The recreation of photos using fMRI scans improved a lot with the use of the new AI system, although the images were less accurate.
The goal behind these studies is to build brain implants that can restore vision by increasing high-level visual processing areas.
This breakthrough opens doors for more advanced solutions and applications in medical technology. It also gives some signs of hope for those who suffer from different types of disability.
Inside Telecom provides you with an extensive list of content covering all aspects of the tech industry. Keep an eye on our Intelligent Tech sections to stay informed and up-to-date with our daily articles.