Unified Multimodal Brain Decoding via Cross-Subject Soft-ROI Fusion

TL;DR

This paper introduces a novel multimodal brain decoding model that enhances cross-subject generalization and interpretability, achieving state-of-the-art results in brain-captioning tasks using fMRI data.

Contribution

The paper proposes a new BrainROI model with a shared soft-ROI space, voxel-gate fusion, and an interpretable prompt optimization process for improved brain decoding.

Findings

Achieved leading results on NSD dataset with improved BLEU-4 and CIDEr scores.

Designed a voxel-wise gated fusion mechanism for better cross-subject transferability.

Implemented an interpretable prompt optimization process that enhances stability and transparency.

Abstract

Multimodal brain decoding aims to reconstruct semantic information that is consistent with visual stimuli from brain activity signals such as fMRI, and then generate readable natural language descriptions. However, multimodal brain decoding still faces key challenges in cross-subject generalization and interpretability. We propose a BrainROI model and achieve leading-level results in brain-captioning evaluation on the NSD dataset. Under the cross-subject setting, compared with recent state-of-the-art methods and representative baselines, metrics such as BLEU-4 and CIDEr show clear improvements. Firstly, to address the heterogeneity of functional brain topology across subjects, we design a new fMRI encoder. We use multi-atlas soft functional parcellations (soft-ROI) as a shared space. We extend the discrete ROI Concatenation strategy in MINDLLM to a voxel-wise gated fusion mechanism (Voxel-gate). We also ensure consistent ROI mapping through global label alignment, which enhances cross-subject transferability. Secondly, to overcome the limitations of manual and black-box prompting methods in stability and transparency, we introduce an interpretable prompt optimization process. In a small-sample closed loop, we use a locally deployed Qwen model to iteratively generate and select human-readable prompts. This process improves the stability of prompt design and preserves an auditable optimization trajectory. Finally, we impose parameterized decoding constraints during inference to further improve the stability and quality of the generated descriptions.