Fig. 2.

: An overview of the dual regression approach utilised within conventional RSN and REACT analyses. Example templates of molecular systems from PET are shown in the top left, analogous to the RSN templates derived from independent components analyses shown in the bottom left. These can be vectorised and entered into the same two-step multiple linear regression framework as shown on the right. In the example of REACT, during stage 1 PET receptor density information is used as a spatial matrix and regressed against the whole brain fMRI data to generate the temporal dynamics associated with each molecular system. In stage 2, these temporal dynamics are then regressed against the BOLD time series in each voxel to generate molecular-enriched FC maps. Finally, these maps can be compared across different experimental conditions such as drug vs placebo. NAT; noradrenaline transporter. DAT; dopamine transporter. SERT; serotonin transporter. VAChT; vesicular acetylcholine transporter. MGluR5; metabotropic glutamate receptor 5. VIS1; medial visual network. DMN; default mode network. SM; somatomotor network. AUD; auditory network. EC; executive control network. FPN; frontoparietal network.