• LOGIN
    Login with username and password
Repository logo

BORIS Portal

Bern Open Repository and Information System

  • Publications
  • Theses
  • Research Data
  • Projects
  • Organizations
  • Researchers
  • More
  • Collections
  • Statistics
  • LOGIN
    Login with username and password
Repository logo
Unibern.ch
  1. Home
  2. Publications
  3. Transformer-based deep learning model for predicting fNIRS short-channel signals.

Transformer-based deep learning model for predicting fNIRS short-channel signals.

Details
Files
DOI
10.48620/92485
Publisher DOI
10.1117/1.NPh.12.4.045008
PubMed ID
41245625
Abstract
Significance
Functional near-infrared spectroscopy (fNIRS) enables portable and noninvasive monitoring of cerebral hemodynamics, but hemodynamic changes originating from extracerebral tissues may influence the signals. To avoid this, short-channel regression (SCR) is widely used, yet physical short-separation detectors are not always available or optimally positioned due to hardware limitations or the experimental setup. In such cases, a virtual, data-driven alternative to physical short-channel detectors may be a viable solution.
Aim
We aimed to (i) develop a transformer-based deep learning model to predict short-separation optical density (OD) signals from long-separation channels and (ii) evaluate whether these virtual signals enable effective SCR.
Approach
We trained the model on a resting-state fNIRS dataset (69 subjects) with paired short- and long-separation recordings. Dual-wavelength OD signals in segmented time windows were used as input for a transformer encoder trained to reconstruct the extracerebral hemodynamic component measured by short channels. Model performance was evaluated using 3 independent datasets: a holdout subset of the same resting-state dataset (23 subjects), a second dataset acquired using a different system (40 subjects), and a task-based finger-tapping dataset (4 subjects). A wavelet coherence-based channel rejection step was optionally applied during preprocessing. Predictions were evaluated using signal similarity metrics (mean squared error [MSE], normalized MSE [NMSE], and Pearson correlation [ r ]) and denoising efficacy (residual variance after regression).
Results
Predicted short-channel signals showed high correspondence with ground-truth measurements in OD (median r = 0.70 and NMSE = 0.047 ) and concentration data (up to r = 0.67 ). When used for SCR, virtual regressors effectively denoise long-channel data. Performance was robust across all datasets, with greater accuracy when low-coherence channels were excluded. In motor task blocks, predicted regressors preserved task-evoked activations and reduced residual variance.
Conclusion
Transformer-based models accurately reconstruct extracerebral hemodynamic signals from long-separation fNIRS data, providing a virtual alternative to physical short channels and supporting standardized, hardware-independent preprocessing.
Date Issued
2025-10
Publication Type
Article
Subject(s)
600 Technology > 610 Medicine & health
Subjects
deep learning
•
functional near-infrared spectroscopy
•
functional near-infrared spectroscopy preprocessing
•
physiological noise
•
short-channel regression
•
signal denoising
•
transformer encoder
Language(s)
en
Author(s)
Guglielmini, Sabino
Banchieri, Vittoria
Scholkmann, Felix  
Institute of Complementary and Integrative Medicine, Anthroposophically Extended Medicine (AeM)  
Institute of Complementary and Integrative Medicine (IKIM)  
Wolf, Martin  
Additional Credits
Institute of Complementary and Integrative Medicine, Anthroposophically Extended Medicine (AeM)  
Institute of Complementary and Integrative Medicine (IKIM)  
Journal
Neurophotonics
Publisher
Society of Photo-optical Instrumentation Engineers
ISSN
2329-423X
Access(Rights)
open.access
Show full item
BORIS Portal
Bern Open Repository and Information System
Build: 0eaa7c [ 7.08. 11:06]
Explore
  • Projects
  • Funding
  • Publications
  • Research Data
  • Organizations
  • Researchers
  • Audiovisual Material
  • Software & other digital items
  • Events
More
  • About BORIS Portal
  • BORIS Portal & Open Science
  • Send Feedback
  • Cookie settings
  • Service Policy
Follow us on
  • Mastodon
  • YouTube
  • LinkedIn
UniBe logo
Repository logo COAR Notify