Fig. 1: Neuroreceptor-based framework can reconstruct the BOLD signal and functional connectivity.

A Neuroreceptor-based framework schematic. The BOLD signal was reconstructed from a linear summation of 19 neuroreceptors and transporters (R₁–R₁₉), each weighted by a corresponding coefficients (β₁–β₁₉). For visualization purposes, we have omitted the constant term in the model. B Reconstruction accuracy of the BOLD signal. Accuracy of neuroreceptor-based framework in reconstructing the BOLD signal over time in a representative participant. Pearson correlation was used to estimate the accuracy between the reconstructed and the empirical brain maps. Brain plots depict the empirical, reconstructed brain and the accuracy between brain maps for representative time points. Line plots depict accuracy over time. Blue line reflects the observed accuracy of the neuroreceptor-based model. Dark and light green lines reflect the accuracy from spatially permuted (spatial) and spin-permutation (spin) null models. C Group level reconstruction accuracy for each of the 1200 time points per participant. D Neuroreceptor-based reconstruction is significantly greater than spatial and spin null models. Bar plots show the average accuracy at the group-level and for the spatial and spin null models. Error bars show mean ± sem. E Reconstructing functional connectivity. Accuracy of reconstructing the group-level functional connectivity. The BOLD signal was concatenated across participants, and Pearson correlation was used to estimate the empirical functional connectivity (left). The reconstructed BOLD signal was concatenated across participants and Pearson correlation was used to estimate the functional connectivity (right). F Comparison with spatial and spin null models. Blue line depicts the accuracy value from panel D. For both null models, the functional connectivity was estimated from the reconstructed BOLD signal by first concatenating the signal across participants, followed by Pearson correlation. ***P < 0.001.