Skip to main content
Fig. 4 | Brain Informatics

Fig. 4

From: HAFNI-enabled largescale platform for neuroimaging informatics (HELPNI)

Fig. 4

The computational pipeline of sparse representation of whole-brain fMRI signals using an online dictionary learning approach. a The whole-brain fMRI signals are aggregated into a big data matrix, in which each row represents the whole-brain fMRI BOLD data in one time point and each column contains the time series of one single voxel. b The target optimization function of dictionary learning and sparse coding. c Illustration of the learned atomic dictionary, each dictionary represents one functional network component. d The coefficient matrix, each row in the matrix measures the weight coefficient of the corresponding dictionary over the whole brain. That is, each row defines the contribution of one dictionary to the composition of all voxel-wise fMRI signals

Back to article page