Fig. 3.

Results of the stochastic HMM inference on resting-state fMRI data from 820 HCP subjects, using a Gaussian distribution to describe each state. (a) Mean activation for three example states. (b) Histogram of the maximum fractional occupancy, a measure aiming to check whether the HMM is able to characterise the dynamics of the data (see Results). (c) Correlation of the state time courses across different runs of the algorithm, showing that the results are robust and consistent across runs. (d) Correlation of the activation maps and functional connectivity between estimations obtained from separate half-splits of the data set, averaged across 5 random splits, and with the states ordered from less to more correlated. (e) Fractional occupancy (defined as the total time spent by the subjects in each state) and distribution of dwell time (i.e. the time spent in each state visit) per state, reflecting basic aspects of the temporal dynamics of the data. (f) Transition probability matrix, reflecting the probability to transition between every pair of states.