Brain Network Dynamics using Deep Learning
DEEP LEARNING MODELS
We develop models using tools from deep learning, such as recurrent neural networks (RNNs). These models can overcome some of the limitations of Hidden Markov Models (HMMs), such as the short memory (the Markovian constraint) and mutual exclusivity of states.
A model we are developing is DYnamic NEtwork Modelling (DYNEMO). This model is inspired by a popular deep learning framework known as the variational autoencoder. Dynemo uses amortised inference to learn a hidden state description of neuroimaging data. Each state represents a large-scale brain network. The temporal dynamics of state switches are captured by an RNN [2].
Tools for performing this kind of analyses are available in the osl-dynamics Python toolbox.
References
- Pervaiz U, Vidaurre D, Gohil C, Smith S, Woolrich M. Multi-dynamic Modelling Reveals Strongly Time-varying Resting fMRI Correlations. Medical image analysis 2022. https://doi.org/10.1016/j.media.2022.102366
- Chetan Gohil, Evan Roberts, Ryan Timms, Alex Skates, Cameron Higgins, Andrew Quinn, Usama Pervaiz, Joost van Amersfoort, Pascal Notin, Yarin Gal, Stanislaw Adaszewski, Mark Woolrich. Mixtures of large-scale dynamic functional brain network modes. bioRxiv 2022.05.03.490453; doi: https://doi.org/10.1101/2022.05.03.490453