Related Experiment Video
Updated: Oct 9, 2025
07:43
Laser-scanning Photostimulation of Optogenetically Targeted Forebrain Circuits
Published on: December 27, 2013
9.4K
NeuroGen: Activation optimized image synthesis for discovery neuroscience
Zijin Gu1, Keith Wakefield Jamison2, Meenakshi Khosla1
1School of Electrical and Computer Engineering, Cornell University, Ithaca, New York, USA.
Neuroimage
|December 22, 2021
Summary
NeuroGen, a new computational strategy, synthesizes images to target brain activity patterns. This tool enhances discovery in human vision neuroscience by reducing noise and amplifying brain response differences.
Area of Science:
- Neuroscience
- Computational Neuroscience
- Artificial Intelligence
Background:
- Functional MRI (fMRI) studies visual cortex responses but faces limitations like noise and limited stimuli.
- Current methods rely on pre-defined hypotheses and observed responses, which can vary significantly between individuals.
Purpose of the Study:
- To introduce NeuroGen, a novel computational strategy for advancing human vision neuroscience discovery.
- To overcome limitations of traditional fMRI experiments by synthesizing targeted visual stimuli.
Main Methods:
- NeuroGen integrates an fMRI-trained neural encoding model with a deep generative network.
- It synthesizes high-fidelity images predicted to elicit specific patterns of brain activation.
Main Results:
- NeuroGen effectively reduces noise and amplifies differences in regional and individual brain responses.
- Synthetic images revealed response patterns not found in natural images, verified by fMRI data.
Conclusions:
- NeuroGen provides a robust architecture for visual neuroscience discovery, extending brain encoding model utility.
- This framework offers a new method for exploring and potentially controlling the human visual system.
