Researchers achieve real-time visual perception reconstruction from fMRI using MindEye2
Decodes brain activity into images within seconds using cloud-based AI pipeline.
Deep Dive
Researchers adapted the MindEye2 fMRI decoding pipeline for real-time use, running it on RT-Cloud to reconstruct perceived natural images from single-trial brain scans in seconds. This proof-of-concept shows that powerful AI decoding can fit within the tight timing constraints of real-time fMRI, paving the way for brain-computer interfaces in scientific discovery and clinical treatment.
Key Points
- Adapted MindEye2, a state-of-the-art fMRI decoder, to run in real time on RT-Cloud, reconstructing perceived images from single-trial brain scans.
- Achieved sub-minute decoding latency, allowing reconstruction before the next stimulus is shown to the participant.
- Open-source platform (RT-Cloud) enables widespread adoption for closed-loop neurofeedback and BCI research.
Why It Matters
Real-time brain decoding could unlock new therapies for mental health and accelerate brain-computer interface development.