Transfer Learning from ImageNet for MEG-Based Decoding of Imagined Speech

TL;DR

This paper demonstrates that transforming MEG signals into image-like representations and applying pretrained vision models significantly improves decoding of imagined speech, achieving high accuracy and revealing shared neural patterns.

Contribution

It introduces an innovative image-based approach for MEG decoding of imagined speech using pretrained vision models, outperforming classical methods.

Findings

Achieved up to 90.4% accuracy in imagined speech detection.

Pretrained vision models capture shared neural representations across subjects.

Temporal analysis localized discriminative information to specific intervals.

Abstract

Non-invasive decoding of imagined speech remains challenging due to weak, distributed signals and limited labeled data. Our paper introduces an image-based approach that transforms magnetoencephalography (MEG) signals into time-frequency representations compatible with pretrained vision models. MEG data from 21 participants performing imagined speech tasks were projected into three spatial scalogram mixtures via a learnable sensor-space convolution, producing compact image-like inputs for ImageNet-pretrained vision architectures. These models outperformed classical and non-pretrained models, achieving up to 90.4% balanced accuracy for imagery vs. silence, 81.0% vs. silent reading, and 60.6% for vowel decoding. Cross-subject evaluation confirmed that pretrained models capture shared neural representations, and temporal analyses localized discriminative information to imagery-locked intervals. These findings show that pretrained vision models applied to image-based MEG representations can effectively capture the structure of imagined speech in non-invasive neural signals.