In this paper, a modified permutation entropy(mPe) algorithm is used to research the coupling relationship between different channels in the Magnetoencephalogram(MEG) signals. We record MEG signals from nine healthy subjects and eight patients with depression who are stimulated by positive and negative emotional images, the mPe values of each channel were calculated separately. The result shows that the patients with depression and healthy subjects had significant differences in the symmetrical regions of the brain, which are the left parietal lobe and the right parietal lobe both positive and negative emotional stimuli. In general, the two symmetric regions of depression patients are more relevant. At the same time, the mPe for the study of MEG can discriminate the diversity between healthy samples and case samples, which has important research signification for the evaluation and diagnosis of clinical pathology.