Frequency-specific dual-attention based adversarial network for blood oxygen level-dependent time series prediction

被引:0
作者
Zheng, Weihao [1 ]
Bao, Cong [1 ]
Mu, Renhui [1 ]
Wang, Jun [2 ,3 ]
Li, Tongtong [1 ]
Zhao, Ziyang [1 ]
Yao, Zhijun [1 ]
Hu, Bin [1 ,4 ,5 ,6 ,7 ]
机构
[1] Lanzhou Univ, Sch Informat Sci & Engn, Gansu Prov Key Lab Wearable Comp, Lanzhou, Peoples R China
[2] Lanzhou Univ, Clin Sch 2, Lanzhou, Peoples R China
[3] Lanzhou Univ, Dept Magnet Resonance, Hosp 2, Lanzhou, Peoples R China
[4] Beijing Inst Technol, Sch Med Technol, Beijing, Peoples R China
[5] Chinese Acad Sci, Shanghai Inst Biol Sci, CAS Ctr Excellence Brain Sci & Intelligence Techno, Shanghai, Peoples R China
[6] Lanzhou Univ, Joint Res Ctr Cognit Neurosensor Technol, Lanzhou, Peoples R China
[7] Chinese Acad Sci, Inst Semicond, Lanzhou, Peoples R China
关键词
autism spectrum disorder; blood oxygen level-dependent (BOLD) series prediction; diagnosis; functional magnetic resonance imaging (fMRI); generative adversarial network; major depressive disorder; FUNCTIONAL CONNECTIVITY; DEFAULT-MODE; FMRI SIGNAL; PARCELLATION; LONG;
D O I
10.1002/hbm.70032
中图分类号
Q189 [神经科学];
学科分类号
071006 ;
摘要
Functional magnetic resonance imaging (fMRI) is currently one of the most popular technologies for measuring brain activity in both research and clinical contexts. However, clinical constraints often result in short fMRI scan durations, limiting the diagnostic performance for brain disorders. To address this limitation, we developed an end-to-end frequency-specific dual-attention-based adversarial network (FDAA-Net) to extend the time series of existing blood oxygen level-dependent (BOLD) data, enhancing their diagnostic utility. Our approach leverages the frequency-dependent nature of fMRI signals using variational mode decomposition (VMD), which adaptively tracks brain activity across different frequency bands. We integrated the generative adversarial network (GAN) with a spatial-temporal attention mechanism to fully capture relationships among spatially distributed brain regions and temporally continuous time windows. We also introduced a novel loss function to estimate the upward and downward trends of each frequency component. We validated FDAA-Net on the Human Connectome Project (HCP) database by comparing the original and predicted time series of brain regions in the default mode network (DMN), a key network activated during rest. FDAA-Net effectively overcame linear frequency-specific challenges and outperformed other popular prediction models. Test-retest reliability experiments demonstrated high consistency between the functional connectivity of predicted outcomes and targets. Furthermore, we examined the clinical applicability of FDAA-Net using short-term fMRI data from individuals with autism spectrum disorder (ASD) and major depressive disorder (MDD). The model achieved a maximum predicted sequence length of 40% of the original scan durations. The prolonged time series improved diagnostic performance by 8.0% for ASD and 11.3% for MDD compared with the original sequences. These findings highlight the potential of fMRI time series prediction to enhance diagnostic power of brain disorders in short fMRI scans. BOLD time series were extracted from preprocessed fMRI data and decomposed into intrinsic mode functions (IMFs) using VMD. Prediction was performed on the IMFs using an adversarial approach and a hybrid loss function. The predicted IMFs were inversely transformed and merged to obtain the final signal sequence of the target region. image
引用
收藏
页数:19
相关论文
共 70 条
  • [1] Anatomical and Functional Assemblies of Brain BOLD Oscillations
    Baria, Alexis T.
    Baliki, Marwan N.
    Parrish, Todd
    Apkarian, A. Vania
    [J]. JOURNAL OF NEUROSCIENCE, 2011, 31 (21) : 7910 - 7919
  • [2] Bedel H.A., 2023, arXiv
  • [3] BolT: Fused window transformers for fMRI time series analysis
    Bedel, Hasan A.
    Sivgin, Irmak
    Dalmaz, Onat
    Dar, Salman U. H.
    Cukur, Tolga
    [J]. MEDICAL IMAGE ANALYSIS, 2023, 88
  • [4] A review and comparison of strategies for multi-step ahead time series forecasting based on the NN5 forecasting competition
    Ben Taieb, Souhaib
    Bontempi, Gianluca
    Atiya, Amir F.
    Sorjamaa, Antti
    [J]. EXPERT SYSTEMS WITH APPLICATIONS, 2012, 39 (08) : 7067 - 7083
  • [5] Default-mode brain dysfunction in mental disorders: A systematic review
    Broyd, Samantha J.
    Demanuele, Charmaine
    Debener, Stefan
    Helps, Suzannah K.
    James, Christopher J.
    Sonuga-Barke, Edmund J. S.
    [J]. NEUROSCIENCE AND BIOBEHAVIORAL REVIEWS, 2009, 33 (03) : 279 - 296
  • [6] Fast and Accurate Time Series Classification Through Supervised Interval Search
    Cabello, Nestor
    Naghizade, Elham
    Qi, Jianzhong
    Kulik, Lars
    [J]. 20TH IEEE INTERNATIONAL CONFERENCE ON DATA MINING (ICDM 2020), 2020, : 948 - 953
  • [7] Evolving Healthcare Delivery in Neurology During the Coronavirus Disease 2019 (COVID-19) Pandemic
    Chen, Patrick M.
    Hemmen, Thomas M.
    [J]. FRONTIERS IN NEUROLOGY, 2020, 11
  • [8] Time series forecasting for nonlinear and non-stationary processes: a review and comparative study
    Cheng, Changqing
    Sa-Ngasoongsong, Akkarapol
    Beyca, Omer
    Trung Le
    Yang, Hui
    Kong, Zhenyu
    Bukkapatnam, Satish T. S.
    [J]. IIE TRANSACTIONS, 2015, 47 (10) : 1053 - 1071
  • [9] Endpoint prediction of heart failure using electronic health records
    Chu, Jiebin
    Dong, Wei
    Huang, Zhengxing
    [J]. JOURNAL OF BIOMEDICAL INFORMATICS, 2020, 109 (109)
  • [10] Cicchetti D. V, 1994, Psychological Assessment, V6, P284, DOI [DOI 10.1037/1040-3590.6.4.284, 10.1037/1040-3590.6.4.284]