ARTFEED — Contemporary Art Intelligence

Real-Time fMRI Decoding of Visual Perception Achieved with MindEye2

ai-technology · 2026-07-29

Researchers have developed a real-time compatible adaptation of the MindEye2 pipeline for reconstructing perceived natural images from fMRI data. Using RT-Cloud, an open-source cloud-based platform, they decoded single-trial visual perception within seconds of image presentation. This overcomes computational constraints that previously prevented state-of-the-art decoding methods from being used in real-time closed-loop neurofeedback. The work demonstrates that fine-grained decoding is achievable in real-time settings, potentially advancing both scientific research and clinical applications of fMRI neurofeedback.

Key facts

  • Real-time closed-loop neurofeedback based on fMRI has led to scientific and clinical advances.
  • Sophisticated analysis methods lag behind state-of-the-art fMRI decoding due to computational factors.
  • Most advanced decoding pipelines do not fit within real-time processing constraints.
  • MindEye2 is a state-of-the-art pipeline for reconstructing perceived natural images.
  • The researchers adapted MindEye2 for real-time compatibility.
  • RT-Cloud is an open-source, scalable cloud-based platform used for the real-time scan.
  • Single-trial visual perception was decoded within seconds after an image was presented.
  • Reliable fine-grained decoding is achievable in real-time settings.

Entities

Sources