fNIRSBCI / neurofeedback

Real-Time Brain Imaging for Thought Control and Neurofeedback

A new deep-learning-powered processing platform cleans wearable HD-DOT data in real time and turns it into 3D brain activity, opening the way for brain–computer interfaces.

21 June 20261 min read

Cleaning a signal corrupted by motion (Contaminated): the conventional TDDR method, this study’s deep-learning method (DAE) and the clean reference (Clean). The brain maps and haemodynamic responses obtained with DAE come closest to the clean reference. Image: Xia et al. (2025), IEEE Trans. Neural Syst. Rehabil. Eng. 33:1220–1230, open access (CC BY) — DOI: 10.1109/TNSRE.2025.3553794.

Brain–computer interfaces (BCI) and neurofeedback require brain activity to be read instantly. This study presents a processing platform that brings wearable optical brain imaging into this “real-time” world.

Why fNIRS/DOT?

BCI and neurofeedback are used more and more in rehabilitation, assistive technologies, neurological diseases and behavioural disorders. fNIRS and DOT are promising for these applications: they are non-invasive, portable and low-cost, with relatively high spatial resolution.

The real challenge: real time

Processing fNIRS/DOT data instantly is not easy. Three things have to be done at once: establishing the measurement’s baseline, correcting motion artefacts (MA) simultaneously across all channels, and (in DOT) speeding up the time-consuming 3D image reconstruction.

The solution

The proposed system combines three components: baseline calibration; a motion-artefact correction model based on a denoising autoencoder (DAE) with a sliding-window strategy; and streamlined reconstruction of 3D brain haemodynamics using a precomputed inverse Jacobian matrix. The data were collected with Gowerlabs LUMO (wearable HD-DOT), and the system was compared with established MA correction methods (MARA, tPCA, wavelet, splineSG, TDDR).

Why does it matter?

This platform makes fNIRS/DOT practical for real-time BCI and neurofeedback. In other words, it opens a concrete door to thought control, rehabilitation with instant brain feedback, and assistive technology applications.