Skip to main content

Federated Multi-trajectory GNNs Under Data Limitations for Baby Brain Connectivity Forecasting

  • Conference paper
Predictive Intelligence in Medicine (PRIME 2023)

Part of the book series: Lecture Notes in Computer Science ((LNCS,volume 14277))

Included in the following conference series:

  • 518 Accesses

Abstract

Building accurate predictive models to forecast the trajectory evolution of baby brain networks during the first postnatal year can provide valuable insights into the dynamics of early brain connectivity development. While emerging studies aimed to predict the evolution of brain graphs from a single observation, they suffer from two major limitations: (i) they typically rely on large training datasets to achieve satisfactory performance. However, longitudinal infant brain scans are costly and hard to acquire, and (ii) they adopt a uni-trajectory approach, lacking the ability to generalize to multi-trajectory prediction tasks, where each graph trajectory corresponds to a particular imaging modality (e.g., functional) and at a fixed resolution (graph size). To address these limitations, we propose FedGmTE-Net*, a federated graph-based multi-trajectory evolution network. Given a small dataset, we leverage the power of federation through collaborative model sharing among diverse hospitals. This approach not only enhances the performance of the local generative graph neural network (GNN) model of each hospital but also ensures the preservation of data privacy. To the best of our knowledge, our framework is the first federated learning framework designed for brain multi-trajectory evolution prediction. Further, to make the most of the limited data available at each hospital, we incorporate an auxiliary regularizer that modifies the local objective function, for more effective utilization of all the longitudinal brain connectivity in the evolution trajectory. This significantly improves the network performance. Our comprehensive experimental results demonstrate that our proposed FedGmTE-Net* outperforms benchmark methods by a substantial margin. Our source code is available at https://github.com/basiralab/FedGmTE-Net.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Subscribe and save

Springer+ Basic
EUR 32.99 /Month
  • Get 10 units per month
  • Download Article/Chapter or Ebook
  • 1 Unit = 1 Article or 1 Chapter
  • Cancel anytime
Subscribe now

Buy Now

Chapter
USD 29.95
Price excludes VAT (USA)
  • Available as PDF
  • Read on any device
  • Instant download
  • Own it forever
eBook
USD 84.99
Price excludes VAT (USA)
  • Available as EPUB and PDF
  • Read on any device
  • Instant download
  • Own it forever
Softcover Book
USD 109.99
Price excludes VAT (USA)
  • Compact, lightweight edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

References

  1. Zhang, H., Shen, D., Lin, W.: Resting-state functional MRI studies on infant brains: a decade of gap-filling efforts. Neuroimage 185, 664–684 (2019)

    Article  Google Scholar 

  2. Stoessl, A.J.: Neuroimaging in the early diagnosis of neurodegenerative disease. Transl. Neurodegeneration 1, 1–6 (2012)

    Article  Google Scholar 

  3. Rekik, I., Li, G., Yap, P.T., Chen, G., Lin, W., Shen, D.: Joint prediction of longitudinal development of cortical surfaces and white matter fibers from neonatal MRI. NeuroImage 152, 411–424 (2017)

    Article  Google Scholar 

  4. Rekik, I., Li, G., Lin, W., Shen, D.: Predicting infant cortical surface development using a 4D varifold-based learning framework and local topography-based shape morphing. Med. Image Anal. 28, 1–12 (2016)

    Article  Google Scholar 

  5. Liu, T.: A few thoughts on brain ROIs. Brain Imaging Behav. 5, 189–202 (2011)

    Article  Google Scholar 

  6. Tzourio-Mazoyer, N., et al.: Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage 15, 273–289 (2002)

    Article  Google Scholar 

  7. Veličković, P.: Everything is connected: graph neural networks. Curr. Opin. Struct. Biol. 79, 102538 (2023)

    Article  Google Scholar 

  8. Zhou, J., et al.: Graph neural networks: a review of methods and applications. AI Open 1, 57–81 (2020)

    Article  Google Scholar 

  9. Bessadok, A., Mahjoub, M.A., Rekik, I.: Graph neural networks in network neuroscience. IEEE Trans. Pattern Anal. Mach. Intell. 45, 5833–5848 (2022)

    Article  Google Scholar 

  10. Göktaş, A.S., Bessadok, A., Rekik, I.: Residual embedding similarity-based network selection for predicting brain network evolution trajectory from a single observation. In: Rekik, I., Adeli, E., Park, S.H., Valdés Hernández, M.C. (eds.) PRIME 2020. LNCS, vol. 12329, pp. 12–23. Springer, Cham (2020). https://doi.org/10.1007/978-3-030-59354-4_2

    Chapter  Google Scholar 

  11. Nebli, A., Kaplan, U.A., Rekik, I.: Deep EvoGraphNet architecture for time-dependent brain graph data synthesis from a single timepoint. In: Rekik, I., Adeli, E., Park, S.H., Valdés Hernández, M.C. (eds.) PRIME 2020. LNCS, vol. 12329, pp. 144–155. Springer, Cham (2020). https://doi.org/10.1007/978-3-030-59354-4_14

    Chapter  Google Scholar 

  12. Tekin, A., Nebli, A., Rekik, I.: Recurrent brain graph mapper for predicting time-dependent brain graph evaluation trajectory. In: Albarqouni, S., et al. (eds.) DART/FAIR -2021. LNCS, vol. 12968, pp. 180–190. Springer, Cham (2021). https://doi.org/10.1007/978-3-030-87722-4_17

    Chapter  Google Scholar 

  13. Kipf, T.N., Welling, M.: Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907 (2016)

  14. Wang, H., et al.: GraphGAN: graph representation learning with generative adversarial nets. In: Proceedings of the AAAI Conference on Artificial Intelligence. Vol. 32 (2018)

    Google Scholar 

  15. Ghribi, O., Li, G., Lin, W., Shen, D., Rekik, I.: Multi-regression based supervised sample selection for predicting baby connectome evolution trajectory from neonatal timepoint. Med. Image Anal. 68, 101853 (2021)

    Article  Google Scholar 

  16. Hestness, J., et al.: Deep learning scaling is predictable, empirically. arXiv preprint arXiv:1712.00409 (2017)

  17. Usman, K., Rajpoot, K.: Brain tumor classification from multi-modality MRI using wavelets and machine learning. Pattern Anal. Appl. 20, 871–881 (2017)

    Article  MathSciNet  Google Scholar 

  18. Bessadok, A., et al.: A few-shot learning graph multi-trajectory evolution network for forecasting multimodal baby connectivity development from a baseline timepoint. In: Rekik, I., Adeli, E., Park, S.H., Schnabel, J. (eds.) PRIME 2021. LNCS, vol. 12928, pp. 11–24. Springer, Cham (2021). https://doi.org/10.1007/978-3-030-87602-9_2

    Chapter  Google Scholar 

  19. Hinton, G., Vinyals, O., Dean, J.: Distilling the knowledge in a neural network. arXiv preprint arXiv:1503.02531 (2015)

  20. McMahan, B., Moore, E., Ramage, D., Hampson, S., y Arcas, B.A.: Communication-efficient learning of deep networks from decentralized data. In: Artificial Intelligence and Statistics, PMLR, pp. 1273–1282 (2017)

    Google Scholar 

  21. Gürler, Z., Nebli, A., Rekik, I.: Foreseeing brain graph evolution over time using deep adversarial network normalizer. In: Rekik, I., Adeli, E., Park, S.H., Valdés Hernández, M.C. (eds.) PRIME 2020. LNCS, vol. 12329, pp. 111–122. Springer, Cham (2020). https://doi.org/10.1007/978-3-030-59354-4_11

    Chapter  Google Scholar 

  22. Gürler, Z., Rekik, I.: Federated brain graph evolution prediction using decentralized connectivity datasets with temporally-varying acquisitions. IEEE Trans. Med. Imaging 42(7), 2022–2031 (2022)

    Google Scholar 

  23. Peterson, L.E.: K-nearest neighbor. Scholarpedia 4, 1883 (2009)

    Article  Google Scholar 

  24. Newman, M.E.: Analysis of weighted networks. Phys. Rev. E 70, 056131 (2004)

    Article  Google Scholar 

Download references

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Islem Rekik .

Editor information

Editors and Affiliations

Rights and permissions

Reprints and permissions

Copyright information

© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG

About this paper

Cite this paper

Pistos, M., Li, G., Lin, W., Shen, D., Rekik, I. (2023). Federated Multi-trajectory GNNs Under Data Limitations for Baby Brain Connectivity Forecasting. In: Rekik, I., Adeli, E., Park, S.H., Cintas, C., Zamzmi, G. (eds) Predictive Intelligence in Medicine. PRIME 2023. Lecture Notes in Computer Science, vol 14277. Springer, Cham. https://doi.org/10.1007/978-3-031-46005-0_11

Download citation

  • DOI: https://doi.org/10.1007/978-3-031-46005-0_11

  • Publisher Name: Springer, Cham

  • Print ISBN: 978-3-031-46004-3

  • Online ISBN: 978-3-031-46005-0

  • eBook Packages: Computer ScienceComputer Science (R0)

Publish with us

Policies and ethics