Decoding fMRI Data into Captions using Prefix Language Modeling

TL;DR

This paper introduces a novel brain decoding method that predicts image embeddings from fMRI signals and uses prefix language modeling to generate captions, reducing computational load and improving accuracy.

Contribution

It proposes a new approach combining DINOv2 embeddings with GPT-2 for captioning from fMRI data, and explores 3D CNNs for better spatial mapping of brain signals.

Findings

Effective decoding of fMRI signals into image captions.

Reduced computational requirements compared to previous methods.

Improved mapping of voxel information using 3D CNNs.

Abstract

With the advancements in Large Language and Latent Diffusion models, brain decoding has achieved remarkable results in recent years. The works on the NSD dataset, with stimuli images from the COCO dataset, leverage the embeddings from the CLIP model for image reconstruction and GIT for captioning. However, the current captioning approach introduces the challenge of potential data contamination given that the GIT model was trained on the COCO dataset. In this work, we present an alternative method for decoding brain signals into image captions by predicting a DINOv2 model's embedding of an image from the corresponding fMRI signal and then providing its [CLS] token as the prefix to the GPT-2 language model which decreases computational requirements considerably. Additionally, instead of commonly used Linear Regression, we explore 3D Convolutional Neural Network mapping of fMRI signals to image embedding space for better accounting positional information of voxels.