The goal of this project is to opensource ongoing efforts to estimate and characterize various aspects of dynamic brain connectivity analysis. As a starting point, we will provide (1a) a set of functions that use scikit-learn, nilearn, bct (+numpy, scipy, nistats,...) that can be used to estimate dynamic connectivity states with a set of connectivity metrics, sliding windows and clustering algorithms, (1b) as well as decomposition methods (e.g. ICA, dictionary learning) (1c) model-based methods (HMM from Vidaurre et al. 2017 PNAS), (2) simple methods to calculate null statistical models using bootstrapping procedures, (3) dynamic graph theory metrics, and (4) visualizations using Spectral Embedding. We encourage interested researchers to join us in order to either help building the examples, or contribute with new methods, or test our methods on their own data.
- sklearn
- nilearn
- matplotlib
- bct
- HMM-MAR from Diego Vidaurre , if you want to run comparisons with HMM. Requires Matlab.
Charles Zheng
Kamil Bonna
Karolina Finc
Xenia Kobeleva
Jessica Dafflon
Sam Guay
Mansoureh Fahimi
Majd Abdallah
Nicolas Farrugia
Julian Kosciessa
- nilearn/plotting script to plot the HMM states
- sklearn script to perform community detection of the intersubject FO correlation matrix (from G_LouvainCommunity.ipynb) and output metastate time-series for each subject -> Fractional Occupancy, switching rate