Diffusion model-based image generation from rat brain activity.

Area of Science:

• Neuroscience
• Computer Science
• Artificial Intelligence
• Art

Background:

• Brain-computer interface (BCI) technology is increasingly used in clinical and assistive applications.
• BCI enables communication, control, and monitoring of cognitive and affective states.
• Current BCI artistic applications often use specific brain signals (e.g., P300, SSVEPs) to control tools, not directly generate images from neural activity.

Purpose of the Study:

• To develop a novel method for generating images directly from continuous brain activity using a latent diffusion model.
• To demonstrate the feasibility of this approach using local field potentials from rat neocortex.
• To explore new possibilities for creative expression and visualization of brain function.

Main Methods:

• Utilized a latent diffusion model, a deep neural network architecture.
• Applied the model to continuous local field potentials recorded from the neocortex of freely moving rats.
• Developed an end-to-end system for real-time image generation from neural data.

Main Results:

• The system successfully converted continuous brain activity into images.
• Generated images reflected the dynamic and stochastic nature of the underlying neural activity.
• Demonstrated a novel procedure for visualizing brain function through direct image generation.

Conclusions:

• The developed BCI method offers a new paradigm for creating art and visualizing brain activity.
• This approach moves beyond controlling tools to directly translating neural signals into visual output.
• The findings open new possibilities for human-computer interaction, artistic expression, and neuroscience research.