Skip to main content
[Preprint]. 2024 Oct 22:2023.09.17.558092. Originally published 2023 Sep 17. [Version 2] doi: 10.1101/2023.09.17.558092

Figure 14: UKB neuroimaging data: Within-modal RDCs (rows I-II and IV-V) and cross-modal RDCs (rows III and VI) of the recovered sources before applying post-hoc CCA.

Figure 14:

Cross-modal subspaces are highlighted in blue while unimodal subspaces are highlighted in green. Within-modal self-correlation patterns show very weak residual dependence between subspaces (rows I-II and IV-V). MSIVA exhibits stronger cross-modal correlations than the unimodal baseline for all predefined subspace structures (row VI vs row III).