Inter-subject modeling of cognitive processes has been a challenging task due to large individual variability in brain structure and function. Graph neural networks (GNNs) provide a potential way to project subject-specific neural responses onto a common representational space by effectively combining local and distributed brain activity through connectome-based constraints. Here we provide in-depth interpretations of biologically-constrained GNNs (BGNNs) that reach state-of-the-art performance in several decoding tasks and reveal inter-subject aligned neural representations underpinning cognitive processes. Specifically, the model not only segregates brain responses at different stages of cognitive tasks, e.g. motor preparation and motor execution, but also uncovers functional gradients in neural representations, e.g. a gradual progression of visual working memory (VWM) from sensory processing to cognitive control and towards behavioral abstraction. Moreover, the multilevel representations of VWM exhibit better inter-subject alignment in brain responses, higher decoding of cognitive states, and strong phenotypic and genetic correlations with individual behavioral performance. Our work demonstrates that biologically constrained deep-learning models have the potential towards both cognitive and biological fidelity in cognitive modeling, and open new avenues to interpretable functional gradients of brain cognition in a wide range of cognitive neuroscience questions.
Background and ObjectivesPathological progression across the cortex is a key feature of Parkinson disease (PD). Cortical gyrification is a morphologic feature of human cerebral cortex that is tightly linked to the integrity of underlying axonal connectivity. Monitoring cortical gyrification reductions may provide a sensitive marker of progression via structural connectivity, preceding the progressive stages of PD pathology. We aimed to examine the progressive cortical gyrification reductions and their associations with overlying cortical thickness, white matter (WM) integrity, striatum dopamine availability, serum neurofilament light (NfL) chain and cerebrospinal fluid (CSF) α-synuclein levels in PD.MethodsThis study included a longitudinal dataset with baseline (T0), 1-year (T1) and 4-year (T4) follow-ups, and two cross-sectional datasets. Local gyrification index (LGI) was computed from T1-weighted MRI data to measure cortical gyrification. Fractional anisotropy (FA) was computed from diffusion-weighted MRI data to measure WM integrity. Striatal binding ratio (SBR) was measured from123Ioflupane SPECT scans. Serum NfL and CSF α-synuclein levels were also measured.ResultsThe longitudinal dataset included 113 de novo PD patients and 55 healthy controls (HC). The cross-sectional datasets included 116 patients with relatively more advanced PD and 85 HC. Compared with HC, de novo PD patients showed accelerated LGI and FA reductions over 1-year period and further decline at 4-year follow-up. Across the three time points, the LGI paralleled and correlated with FA (p=0.002 at T0, p=0.0214 at T1, p=0.0037 at T4) and SBR (p=0.0095 at T0, p=0.0035 at T1, p=0.0096 at T4) but not with overlying cortical thickness in PD patients. Both LGI and FA correlated with serum NfL level (LGI: p<0.0001 at T0, p=0.0043 at T1; FA: p<0.0001 at T0, p=0.0001 at T1) but not with CSF α-synuclein level in PD patients. In the two cross-sectional datasets, we revealed similar patterns of LGI and FA reductions and associations between LGI and FA in patients with more advanced PD.DiscussionWe demonstrated progressive reductions in cortical gyrification that were robustly associated with WM microstructure, striatum dopamine availability and serum NfL level in PD. Our findings may contribute biomarkers for PD progression and potential pathways for early interventions of PD.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.