Visual Imagery Decoding Fmri

---
name: visual-imagery-decoding-fmri
description: Latent functional alignment approach for decoding visual imagery from fMRI data. Extends perception-optimized pipelines to mental imagery using diffusion-based generative models. Activation: fmri, decoding, visual-imagery, diffusion-models, functional-alignment, neuroscience, brain, neural
version: 1.0.0
author: Hermes Agent
license: MIT
metadata:
  hermes:
    source_paper: "Seeing the imagined: a latent functional alignment in visual imagery decoding from fMRI data (arXiv:2604.15374)"
    authors: "Fabrizio Spera, Tommaso Boccato, Michal Olak et al."
    published: 2026-04-15
    tags: ["fmri", "decoding", "visual-imagery", "diffusion-models", "functional-alignment"]
---

# Visual Imagery Decoding from fMRI via Latent Functional Alignment

## Overview
Extends fMRI visual decoding pipelines from perception to mental imagery using latent functional alignment and diffusion-based generative models. Leverages large-scale datasets like NSD.

Based on: [Seeing the imagined: a latent functional alignment in visual imagery decoding from fMRI data](https://arxiv.org/abs/2604.15374) (2026-04-15)

## Key Insights
- Latent functional alignment bridges perception and imagery
- Diffusion-based generative models for reconstruction
- Leverages Natural Scenes Dataset (NSD)
- Performance gap between perception and mental imagery
- Cross-subject generalization via functional alignment

## Applications
- Brain-computer interfaces
- Mental state decoding
- Cognitive neuroscience
- Neuroimaging

## Abstract
We extend fMRI visual decoding pipelines from perception to mental imagery using latent functional alignment and diffusion-based generative models. Our approach leverages the Natural Scenes Dataset (NSD) and demonstrates cross-subject generalization via functional alignment, revealing the performance gap between perception and mental imagery decoding.

## Methodology

### Latent Functional Alignment
- Align subjects in latent space for cross-subject generalization
- Diffusion-based generative models for image reconstruction
- Extends perception decoding to mental imagery

## Reference
Fabrizio Spera, Tommaso Boccato, Michal Olak et al.. Seeing the imagined: a latent functional alignment in visual imagery decoding from fMRI data. arXiv:2604.15374, 2026-04-15.
URL: https://arxiv.org/abs/2604.15374