New framework enhances EEG-to-fNIRS signal generation for brain-computer interfaces

2026-09-25

Researchers have developed Bio-MF, a one-step framework for generating fNIRS signals from EEG data. This aims to improve hybrid motor-imagery brain-computer interfaces by providing hemodynamic information when direct fNIRS acquisition is not feasible.

VERA Brief

AI-generated. Grounded in the article and its cited sources.

Researchers developed Bio-MF, a framework that generates fNIRS signals from EEG data. This aims to improve hybrid brain-computer interfaces by offering hemodynamic information when direct fNIRS acquisition is not possible.

Key facts

Bio-MF is a one-step framework for generating fNIRS signals from EEG data.
The framework is designed to enhance hybrid motor-imagery brain-computer interfaces.
Bio-MF incorporates Spatial-Temporal Interactive 4D Encoding, cross-modal classifier-free guidance, and noise-level-gated FFT regularization.
On Dataset 1, the combination of EEG with synthetic fNIRS improved accuracy by 3.37 percentage points for HbR and 4.15 percentage points for HbO.
On Dataset 2, ACC improvements were 2.98 percentage points for HbR and 2.50 percentage points for HbO.

Source: arXiv · cs.LG

Reported by VERA Newswire.