Segregation of the regional radiomics similarity network exhibited an increase from late childhood to early adolescence: A developmental investigation

• DOI: 10.1016/j.neuroimage.2024.120893
• Publication Date: 2024-10-18T18:11:21Z
• Authors: Lei Chu; Debin Zeng; Yirong He; Xiaoxi Dong; Qiongling Li; Xuhong Liao; Tengda Zhao; Xiaodan Chen; Tianyuan Lei; Weiwei Men; Yanpei Wang; Daoyang Wang; Mingming Hu; Zhiying Pan; Shuping Tan; Jia-Hong Gao; Shaozheng Qin; Sha Tao; Qi Dong; Yong He; Shuyu Li
• Keywords: Similarity (geometry); Late childhood

Brain development is characterized by an increase in structural and functional segregation, which supports the specialization of cognitive processes within the context of network neuroscience. Despite the potential of radiomics features to capture microenvironmental information and lesion homogeneity, our understanding of whether these features can illuminate brain developmental changes remains limited. In this study, we constructed individual Regional Radiomics Similarity Networks (R2SNs) from a longitudinal dataset of 494 T1-weighted MR scans from 309 typically developing children aged 6.2 to 13 years at baseline. Linear mixed-effect models revealed longitudinal increases in R2SN segregation indices, particularly within limbic, dorsal attention network at the system level, and decreases in segregation indices within the ventral attention network. Superior performance in working memory and inhibitory control was associated with higher system-level segregation indices in the default, frontoparietal, ventral attention, somatomotor and subcortical systems, and lower local segregation indices in several regions of the visual network, regardless of age. Furthermore, gene enrichment analysis revealed a correlation between age-related changes in the local segregation index and regional expression levels of genes related to developmental processes. These findings provide novel insights into typical brain developmental changes as indicated by the segregation index computed using R2SN approaches, and can be a valuable tool for understanding human brain structural and cognitive maturation.