Skip to main content
. Author manuscript; available in PMC: 2014 Jun 1.
Published in final edited form as: Biometrics. 2013 Oct 15;69(4):10.1111/biom.12085. doi: 10.1111/biom.12085

Figure 2.

Figure 2

Simulation results on comparison of STM, GMRF with no transformation, and the voxel-wise linear model. Panels (a)–(d) represent the pattern of β used to generate the images; panels (e)–(h) represent the estimated β obtained from the least squares estimator in Matlab; panels (i)–(l) represent the posterior mean of β obtained by fitting a GMRF model with no transformation; and panels (m)–(p) are the posterior mean of β obtained from our STM. The inclusion of the transformation parameter substantially improves the estimation of the true underlying pattern.