Statistical parametric mapping (SPM) depends on the general linear model(GLM) and the theory of Gaussian fields to some extent. But the disadvantages of the GLM are related to the fact that these assumptions outlined do not fairly represent the fMRI data. Also, hemodynamics and distributed patterns of the brain activity may not be well modeled by the GLM regression framework. While, the independent component analysis(ICA) does not provide the investigator with a significance estimate for each component activation, which may discourage experimenters from attempting to interpret the results. The paper proposes a method which combines some of thebenefits of ICA with the hypothesis-testing approach of the SPM. Experimental results demonstrate that the proposed method is effective for detecting the activations resulting from a motor task.