Our paper titled A Unified fNIRS Classification Framework Informed by Local Brain Activation Patterns has been accepted by IEEE Transactions on Industrial Informatics (https://ieeexplore.ieee.org/document/9670659).
🌟 Our other work may also be useful to you：
✅ 🆕 🎯 Domain Knowledge Fused State Space Model for fNIRS-Based Brain–Computer Interfaces (https://ieeexplore.ieee.org/document/11522669, GitHub: https://github.com/wzhlearning/fNIRS4D).
✅ Rethinking Delayed Hemodynamic Responses for fNIRS Classification (https://ieeexplore.ieee.org/document/10311392/, GitHub: https://github.com/wzhlearning/fNIRSNet).
✅ Transformer Model for Functional Near-Infrared Spectroscopy Classification (https://ieeexplore.ieee.org/document/9670659, GitHub: https://github.com/wzhlearning/fNIRS-Transformer).
✅ A General and Scalable Vision Framework for Functional Near-Infrared Spectroscopy Classification ( https://ieeexplore.ieee.org/abstract/document/9828508).
Recent studies have focused on task-specific and neuroscience-agnostic fNIRS classification models rather than a unified neuroscience-informed framework. We propose LoBrAFrame, a unified, neuroscience-informed fNIRS classification framework that leverages local brain activation patterns through a shared weight mechanism. Within this framework, researchers can easily enhance classification performance using simple or off-the-shelf methods, without redesigning complex models. To instantiate a concrete model, we introduce Mamba, a state space model, into the fNIRS domain and propose LoBrAMamba. Our work will inspire interest in neuroscience-informed fNIRS frameworks.
conda create -n lobraframe python=3.9
conda activate lobraframe
conda install pytorch==2.1.1 torchvision==0.16.1 torchaudio==2.1.1 pytorch-cuda=11.8 -c pytorch -c nvidia
conda install -c "nvidia/label/cuda-11.8.0" libcusolver-devYou should install mamba:
pip install causal-conv1d>=1.4.0
pip install mamba-ssm==2.2.1No special requirements exist for other Python libraries needed for operation.
Use the .mat files provided in the scripts folder to preprocess the dataset. You need to download the BBCI library from the GitHub link.
The scripts for MA and UFFT originate from our previous work (fNIRSNet: https://ieeexplore.ieee.org/document/10311392). The GitHub link is https://github.com/wzhlearning/fNIRSNet.
3.1 Manually specify dataset tasks (task_id) and paths (UFFT_data_path, MA_data_path, WG_data_path).
# Select dataset through task_id
task = ['UFFT', 'MA', 'WG']
task_id = 0
# Set dataset path
UFFT_data_path = 'UFFT_data'
MA_data_path = 'MA_fNIRS_data'
WG_data_path = 'WG_fNIRS_data'3.2 Run subject-specific experiments
python KFold_Train.py3.3 Run subject-independent experiments
python LOSO_Train.py4.1 Manually specify dataset tasks (task_id).
# Select dataset through task_id
task = ['UFFT', 'MA', 'WG']
task_id = 04.2 Run subject-specific results
python KFold_Results.py4.3 Run subject-independent results
python LOSO_Results.pyIf you found the study useful for you, please consider citing and star it.
@ARTICLE{Wang2025LoBrAFrame,
author = {Wang, Zenghui and Du, Songlin},
journal = {IEEE Transactions on Industrial Informatics},
title = {A Unified fNIRS Classification Framework Informed by Local Brain Activation Patterns},
year = {2026},
volume = {22},
number = {3},
pages = {1871-1881},
doi = {10.1109/TII.2025.3632147}
}
🌟 Our other works may also be beneficial to you, please consider citing them.
https://ieeexplore.ieee.org/document/11522669
https://github.com/wzhlearning/fNIRS4D
@ARTICLE{Wang2026fNIRS4D,
author = {Wang, Zenghui and Zhu, Delv and Zhang, Jun and Xiao, Guobao and Du, Songlin},
journal = {IEEE Transactions on Instrumentation and Measurement},
title = {Domain Knowledge Fused State Space Model for fNIRS-Based Brain–Computer Interfaces},
year = {2026},
volume = {75},
number = {},
pages = {4008016-4008016},
doi = {10.1109/TIM.2026.3693454}}
https://ieeexplore.ieee.org/document/10311392
https://github.com/wzhlearning/fNIRSNet
@ARTICLE{wang2023fNIRSNet,
author = {Wang, Zenghui and Fang, Jihong and Zhang, Jun},
journal = {IEEE Transactions on Neural Systems and Rehabilitation Engineering},
title = {Rethinking Delayed Hemodynamic Responses for fNIRS Classification},
year = {2023},
volume = {31},
number = {},
pages = {4528-4538},
doi = {10.1109/TNSRE.2023.3330911}}
https://ieeexplore.ieee.org/document/9670659
https://github.com/wzhlearning/fNIRS-Transformer
@ARTICLE{Wang2022Transformer,
author = {Wang, Zenghui and Zhang, Jun and Zhang, Xiaochu and Chen, Peng and Wang, Bing},
journal = {IEEE Journal of Biomedical and Health Informatics},
title = {Transformer Model for Functional Near-Infrared Spectroscopy Classification},
year = {2022},
volume = {26},
number = {6},
pages = {2559-2569},
doi = {10.1109/JBHI.2022.3140531}}
https://ieeexplore.ieee.org/abstract/document/9828508
@ARTICLE{Wang2022VisionfNIRSFramework,
author = {Wang, Zenghui and Zhang, Jun and Xia, Yi and Chen, Peng and Wang, Bing},
journal = {IEEE Transactions on Neural Systems and Rehabilitation Engineering},
title = {A General and Scalable Vision Framework for Functional Near-Infrared Spectroscopy Classification},
year = {2022},
volume = {30},
number = {},
pages = {1982-1991},
doi = {10.1109/TNSRE.2022.3190431}}