EUVIMEDEUVIMED entdecken
Crossref

Spec2VolCAMU-Net: a spectrogram-to-volume model for EEG-to-fMRI reconstruction based on Multi-directional Time–Frequency Convolutional Attention Encoder and Vision-Mamba U-Net

Dongyi He iD, Shiyang Li, Bin Jiang iD, He Yan

Journal of Neural Engineering · 01.10.2025 · 22 (5) · 056042

DOI: 10.1088/1741-2552/ae15bf ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Neural Engineering, online erschienen am 30.10.2025. Verfasst von Dongyi He und 3 weiteren Autor:innen. Beteiligte Einrichtung: School of Artificial Intelligence, Chongqing University of Technology, Chongqing 400054, People's Republic of China. Hauptthemen laut MEDLINE: Magnetic Resonance Imaging, Electroencephalography, Neural Networks, Computer, Brain. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 35 Literaturangaben. Gefördert von: Chongqing Municipal Key Project for Technology Innovation and Application Development, National Natural Science Foundation of P.R. China, Scientific and Technological Research Program of the Chongqing Education Commission.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Objective. High-resolution functional magnetic resonance imaging (fMRI) is essential for mapping human brain activity; however, it remains costly and logistically challenging. If comparable volumes could be generated directly from widely available scalp electroencephalography (EEG), advanced neuroimaging would become significantly more accessible. Existing EEG-to-fMRI generators rely on plain convolutional neural networks that fail to capture cross-channel time–frequency cues or on heavy transformer/generative adversarial network decoders that strain memory and stability. Approach. To address these limitations, we propose Spec2VolCAMU-Net, a lightweight architecture featuring a Multi-directional Time–Frequency Convolutional Attention Encoder for rich feature extraction and a Vision-Mamba U-Net decoder that uses linear-time state-space blocks for efficient long-range spatial modeling. We frame the goal of this work as establishing a new state of the art in the spatial fidelity of single-volume reconstruction, a foundational prerequisite for the ultimate aim of generating temporally coherent fMRI time series. Main results. Trained end-to-end with a hybrid SSI-MSE loss, Spec2VolCAMU-Net achieves state-of-the-art fidelity on three public benchmarks, recording structural similarity index (SSIM) of 0.693 on NODDI, 0.725 on Oddball and 0.788 on CN-EPFL, representing improvements of 14.5%, 14.9%, and 16.9% respectively over previous best SSIM scores. Furthermore, it achieves competitive peak signal-to-noise ratio (PSNR) scores, particularly excelling on the CN-EPFL dataset with a 4.6% improvement over the previous best PSNR, thus striking a better balance in reconstruction quality. Significance. The proposed model is lightweight and efficient, making it suitable for real-time applications in clinical and research settings. The code is available at https://github.com/hdy6438/Spec2VolCAMU-Net .

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Magnetic Resonance ImagingElectroencephalographyNeural Networks, ComputerBrainAttentionBrain MappingImage Processing, Computer-AssistedHumans

Stichwörter der Autor:innen

EEG-to-fMRI synthesisdeep learningelectroencephalography (EEG)functional magnetic resonance imaging (fMRI)

Autor:innen und Institutionen

  • School of Artificial Intelligence, Chongqing University of Technology, Chongqing 400054, People's Republic of China.

ORCID-Profile: Dongyi He, Bin Jiang

Förderung

  • Chongqing Municipal Key Project for Technology Innovation and Application Development
  • National Natural Science Foundation of P.R. China
  • Scientific and Technological Research Program of the Chongqing Education Commission
  • College Student Innovation and Entrepreneurship Project
  • Natural Science Foundation of Chongqing Municipality

Zitieren

APA

He, D., Li, S., Jiang, B., & Yan, H. (2025). Spec2VolCAMU-Net: a spectrogram-to-volume model for EEG-to-fMRI reconstruction based on Multi-directional Time–Frequency Convolutional Attention Encoder and Vision-Mamba U-Net. Journal of Neural Engineering, 22(5), 056042. https://doi.org/10.1088/1741-2552/ae15bf

Vancouver

He D, Li S, Jiang B, Yan H. Spec2VolCAMU-Net: a spectrogram-to-volume model for EEG-to-fMRI reconstruction based on Multi-directional Time–Frequency Convolutional Attention Encoder and Vision-Mamba U-Net. Journal of Neural Engineering. 2025;22(5):056042. doi:10.1088/1741-2552/ae15bf

BibTeX
@article{He2025,
  author = {He, Dongyi and Li, Shiyang and Jiang, Bin and Yan, He},
  title = {Spec2VolCAMU-Net: a spectrogram-to-volume model for EEG-to-fMRI reconstruction based on Multi-directional Time–Frequency Convolutional Attention Encoder and Vision-Mamba U-Net},
  journal = {Journal of Neural Engineering},
  year = {2025},
  volume = {22},
  number = {5},
  pages = {056042},
  doi = {10.1088/1741-2552/ae15bf}
}
Kostenlos starten →