MindEye2: Shared-Subject Models Enable fMRI-To-Image With 1 Hour of Data
This work addresses the practical limitation of expensive fMRI data collection for brain-to-image models, enabling accurate reconstructions from a single MRI visit.
The paper tackles the problem of high fMRI data requirements for visual perception reconstruction by introducing MindEye2, which achieves high-quality reconstructions using only 1 hour of training data per subject, improving out-of-subject generalization and attaining state-of-the-art metrics compared to single-subject approaches.
Reconstructions of visual perception from brain activity have improved tremendously, but the practical utility of such methods has been limited. This is because such models are trained independently per subject where each subject requires dozens of hours of expensive fMRI training data to attain high-quality results. The present work showcases high-quality reconstructions using only 1 hour of fMRI training data. We pretrain our model across 7 subjects and then fine-tune on minimal data from a new subject. Our novel functional alignment procedure linearly maps all brain data to a shared-subject latent space, followed by a shared non-linear mapping to CLIP image space. We then map from CLIP space to pixel space by fine-tuning Stable Diffusion XL to accept CLIP latents as inputs instead of text. This approach improves out-of-subject generalization with limited training data and also attains state-of-the-art image retrieval and reconstruction metrics compared to single-subject approaches. MindEye2 demonstrates how accurate reconstructions of perception are possible from a single visit to the MRI facility. All code is available on GitHub.