Independent component analysis (ICA), as a data-driven signal decomposition method, has been widely used in fMRI. Sources of the measurement can be separated according to the rule of maximum independency, but it usually cannot naturally generate a source which is highly correlated with the signal we are interested in. To solve this problem, we propose a new method, prior knowledge oriented ICA (pICA), to drive ICA to a set of sources with the SOI among them. Experiments of simulation and fMRI show this new method has higher specificity and accuracy in identifying the SOI and its corresponding spatial map.
This abstract and the presentation materials are available to members only; a login is required.