Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Modality-agnostic decoding of vision and language from fMRI.

eLife·2026
Same author

Evidence for compositionality in fMRI visual representations via Brain Algebra.

Communications biology·2025
Same author

Through their eyes: Multi-subject brain decoding with simple alignment techniques.

Imaging neuroscience (Cambridge, Mass.)·2025
Same author

Random Tactile Noise Stimulation Reveals Beta-Rhythmic Impulse Response Function of the Somatosensory System.

The Journal of neuroscience : the official journal of the Society for Neuroscience·2023
Same author

Distinct roles of forward and backward alpha-band waves in spatial visual attention.

eLife·2023
Same author

α Phase-Amplitude Tradeoffs Predict Visual Perception.

eNeuro·2022

Related Experiment Video

Updated: Jul 16, 2025

Co-analysis of Brain Structure and Function using fMRI and Diffusion-weighted Imaging
17:06

Co-analysis of Brain Structure and Function using fMRI and Diffusion-weighted Imaging

Published on: November 8, 2012

26.3K

Natural scene reconstruction from fMRI signals using generative latent diffusion.

Furkan Ozcelik1,2, Rufin VanRullen3,4,5

  • 1CerCo, CNRS UMR5549, Toulouse, France. ozcelikfu@gmail.com.

Scientific Reports
|September 21, 2023
PubMed
Summary

This study introduces Brain-Diffuser, a novel framework for reconstructing natural images from fMRI signals. It effectively combines low-level and high-level visual features for advanced neural decoding.

More Related Videos

Concurrent EEG and Functional MRI Recording and Integration Analysis for Dynamic Cortical Activity Imaging
11:28

Concurrent EEG and Functional MRI Recording and Integration Analysis for Dynamic Cortical Activity Imaging

Published on: June 30, 2018

11.7K
Modeling the Functional Network for Spatial Navigation in the Human Brain
05:55

Modeling the Functional Network for Spatial Navigation in the Human Brain

Published on: October 13, 2023

1.1K

Related Experiment Videos

Last Updated: Jul 16, 2025

Co-analysis of Brain Structure and Function using fMRI and Diffusion-weighted Imaging
17:06

Co-analysis of Brain Structure and Function using fMRI and Diffusion-weighted Imaging

Published on: November 8, 2012

26.3K
Concurrent EEG and Functional MRI Recording and Integration Analysis for Dynamic Cortical Activity Imaging
11:28

Concurrent EEG and Functional MRI Recording and Integration Analysis for Dynamic Cortical Activity Imaging

Published on: June 30, 2018

11.7K
Modeling the Functional Network for Spatial Navigation in the Human Brain
05:55

Modeling the Functional Network for Spatial Navigation in the Human Brain

Published on: October 13, 2023

1.1K

Area of Science:

  • Neuroscience
  • Artificial Intelligence
  • Computer Vision

Background:

  • Neural decoding aims to reconstruct perceived images from brain activity, specifically fMRI signals.
  • Prior methods struggle to jointly reconstruct low-level (shape, texture) and high-level (semantics) visual properties for complex scenes.
  • Generative AI, particularly latent diffusion models, offers new potential for high-complexity image generation in brain decoding.

Purpose of the Study:

  • To develop an advanced framework for reconstructing natural scenes from fMRI signals by leveraging generative AI.
  • To integrate low-level and high-level visual feature reconstruction for more comprehensive image decoding.
  • To evaluate the proposed framework's performance against existing methods and its utility in neuroscience.

Main Methods:

  • A two-stage reconstruction framework, "Brain-Diffuser", was developed.
  • Stage 1: Reconstructs images with low-level properties and layout from fMRI signals using a Very Deep Variational Autoencoder (VDVAE).
  • Stage 2: Employs a latent diffusion model (Versatile Diffusion) for image-to-image generation, conditioned on multimodal (text and visual) features.

Main Results:

  • The Brain-Diffuser framework significantly outperforms previous models on the Natural Scenes Dataset benchmark, both qualitatively and quantitatively.
  • The model successfully generates compelling "ROI-optimal" scenes from synthetic fMRI patterns, aligning with neuroscientific principles.
  • Demonstrates effective reconstruction of complex scene properties by integrating low-level and high-level visual information.

Conclusions:

  • The Brain-Diffuser methodology represents a significant advancement in neural decoding for natural image reconstruction.
  • This approach effectively combines VDVAE and latent diffusion models for enhanced brain decoding capabilities.
  • The framework holds potential for applications in brain-computer interfaces and fundamental neuroscience research.