Signal correlation measure in multi-echo T2*-w MR images

Signal correlation is mainly measured from similarity or dissimilarity between a tissue of interest and a tissue of reference. In this paper, similarity coefficient mapping (SCM) was modified to estimate signal correlation from both aspects and generates two maps. One map (RMM) depicts relative signal magnitude and the other (RM) reveals tissue difference. The modified method was validated with a digital phantom and clinical human brain magnetic resonance images. Experimental results show that (a) both RMM and SCM map comparably improve the perceptibility of anatomical structures; (b) tissue contrast is enhanced in RM that pixel intensities of the reference tissue region are much lower than those of non-reference region; and (c) intra-tissue similarity and inter-tissue dissimilarity can be cross-validated from pixel intensity analysis based on the generated map pair. This method benefits image fusion, tissue classification and malformation detection.