Anatomically constrained squeeze-and-excitation graph attention network for cortical surface parcellation.

In order to understand the organizational structures of healthy cerebral cortex and the abnormalities in neurological and psychiatric diseases, it is significant to parcellate the cortical surface. The cortical surface, however, is a highly folded complex geometric structure which challenges automatic cortical surface parcellation. Nowadays, the parcellation methods of cerebral cortex are mostly based on geometric simplification, i.e., iteratively inflating and mapping the cortical surface to a spherical surface for processing, which is time-consuming and cannot make full use of the intrinsic structural information of the original cortical surface. In this study, we proposed an anatomically constrained squeeze-and-excitation graph attention network (ASEGAT) for an end-to-end brain cortical surface parcellation on the original cortical surface manifold. The ASEGAT is formed by two graph attention modules and a squeeze-and-excitation module that incorporate self-attention and head attention for rendering features of each node. Furthermore, we designed an anatomic constraint loss to introduce the anatomical priori of regional adjacency relationships, which could improve the consistency of region labeling. We evaluated our model on a public dataset of 100 manually labeled brain surfaces. Compared with several advanced methods, the results showed that our proposed approach achieved state-of-the-art performance, obtaining an accuracy of 90.65% and a dice score of 89.00%.

Autism spectrum disorder diagnosis using graph attention network based on spatial-constrained sparse functional brain networks.

The accurate diagnosis of autism spectrum disorder (ASD), a common mental disease in children, has always been an important task in clinical practice. In recent years, the use of graph neural network (GNN) based on functional brain network (FBN) has shown powerful performance for disease diagnosis. The challenge to construct "ideal" FBN from resting-state fMRI data remained. Moreover, it remains unclear whether and to what extent the non-Euclidean structure of different FBNs affect the performance of GNN-based disease classification. In this paper, we proposed a new method named Pearson's correlation-based Spatial Constraints Representation (PSCR) to estimate the FBN structures that were transformed to brain graphs and then fed into a graph attention network (GAT) to diagnose ASD. Extensive experiments on comparing different FBN construction methods and classification frameworks were conducted on the ABIDE I dataset (n = 871). The results demonstrated the superiority of our PSCR method and the influence of different FBNs on the GNN-based classification results. The proposed PSCR and GAT framework achieved promising classification results for ASD (accuracy: 72.40%), which significantly outperformed competing methods. This will help facilitate patient-control separation, and provide a promising solution for future disease diagnosis based on the FBN and GNN framework.

Atypical sulcal pattern in boys with attention-deficit/hyperactivity disorder.

Neurodevelopmental disorders, such as attention-deficit/hyperactivity disorder (ADHD), are often accompanied by disrupted cortical folding. We applied a quantitative sulcal pattern analysis technique using graph structures to study the atypical cortical folding at the lobar level in ADHD brains in this study. A total of 183 ADHD patients and 167 typical developmental controls matched according to age and gender were enrolled. We first constructed sulcal graphs at the brain lobar level and then investigated their similarity to the typical sulcal patterns. The within-group variability and interhemispheric similarity in sulcal patterns were also compared between the ADHD and TDC groups. The results showed that, compared with controls, the left frontal, right parietal, and temporal lobes displayed altered similarities to the typical sulcal patterns in patients with ADHD. Moreover, the sulcal patterns in ADHD seem to be more heterogeneous than those in controls. The results also identified the disruption of the typical asymmetric sulcal patterns in the frontal lobe between the ADHD and control groups. Taken together, our results revealed the atypical sulcal pattern in boys with ADHD and provide new insights into the neuroanatomical mechanisms of ADHD.