Novel approaches to analyzing time series if objects with applications to brain connectivity.

Understanding Human Brain Function as Networks

Different brain regions interact with each other as a network, just like a group of friends in a social network. The concept of modelling connectivity between brain regions has become an important framework to enhance our understanding of human brain’s structure and function.

Yet analysis of brain networks remains very challenging due to complexity of the network data both in space and time, as brain networks tend to change over time. This leads to a new research field - network neuroscience, leveraging on recent developments in statistics, machine learning, information theory, and computational neuroscience, to extract patterns and new insights from brain network data.

This new research has brought together Dr Ting Chee Ming from School of IT, Monash University Malaysia and biostatistician Prof Hernando Ombao from the King Abdullah University of Science and Technology (KAUST), Saudi Arabia.

Ting and Ombao’s groups have been actively collaborating in developing novel statistical and machine learning methods for analysing, modelling and classifying brain connectivity data.

“We have developed statistical data analytic tools to characterize dependence among high-dimensional neuroimaging time series, such as functional magnetic resonance imaging (fMRI) and electroencephalograms (EEG)” says Ting. In collaboration with experimental psychologist Dr Jeremy Skipper from University College London, Ting and Ombao has recently proposed a novel decomposition method for multi-subject brain networks. The method when applied to movie-watching fMRI dataset, uncover the role of speech production networks in perceiving in-coming speech in noisy environment.

In support of their research, Ombao and Ting have been recently awarded a Competitive Research Grant (CRG) from KAUST. Research funded through the CRG program is expected to lead to major discoveries and innovations at the frontiers of science and technology. In this project, Ombao and Ting’s group will devise novel approaches to analyzing time series of objects with applications to brain connectivity.

“Our hope is to find associations between brain connectivity and neuropsychiatric and neurological diseases, which can support clinical diagnosis and treatment” according to Ting. Ting’s group is also currently developing graph deep learning models to predict major depression and individual behaviour from human brain networks. “We have also been collaborating with our hospitals such as the Hospital Canselor Tuanku Muhriz UKM on spectral and connectivity of EEG signals for stroke and epileptic patients. These methods could provide clinically useful tools for prediction of stroke severity and seizure localization” says Ting.

Figure: Difference in EEG-derived brain connectivity patterns between normal and seizure state in an epileptic patient.

References:

  1. Ting, C.-M., Skipper, J.I., Noman, F, Small, S.L. and Ombao, H. (2022). Separating stimulus-induced and background components of dynamic functional connectivity in naturalistic fMRI, IEEE Trans Medical Imaging, 41(6), 1431-1442.
  2. Noman, F., Yap, S.-Y., Phan, R. C. -W., Ting, C.-M. and Ombao, H. (2022) Graph autoencoder-based embedded learning in dynamic brain networks for autism spectrum disorder identification. In Proc. 29th IEEE International Conference on Image Processing.
  3. Ting, C.-M., Samdin, S.B., Tang, M. and Ombao, H. (2021). Detecting dynamic community structure in functional brain networks across individuals: A multilayer approach, IEEE Trans Medical Imaging, 40(2), 468 - 480.
  4. Phang, C.-R., Noman, F., Hussain, H., Ting, C.-M., Ombao, H. (2020). A multi-domain connectome convolutional neural network for identifying schizophrenia from EEG connectivity patterns, IEEE J Biomed Health Infor, 24(5).
  5. Ting, C.-M., Ombao, H., Salleh, Sh-H. and Abd Latif, A.Z. (2020). Multi-scale factor analysis of high-dimensional functional connectivity in brain networks. IEEE Trans. Network Science Eng, 7(1), 449 - 465
  6. Ting, C.-M., Ombao, H., Samdin, S.B. and Salleh, Sh-H. (2018). Estimating dynamic connectivity states in fMRI using regime-switching factor models. IEEE Trans. Medical Imaging, 37(4), 1011-1023.
  7. Ombao, H., Fiecas, M.,Ting C.-M. and Low, Y.F. (2018). Statistical models for brain signals with properties that evolve across trials. NeuroImage, 180, 609-618.
  8. Samdin, S. B., Ting, C.-M., Ombao, H. and Salleh, Sh-H. (2017). A unified estimation framework for state-related changes in effective brain connectivity. IEEE Trans. Biomedical Engineering, 64(4), 844-858.