During early and middle gestation, various populations of subplate neurons control ingrowing of thalamocortical, basal forebrain cholinergic, and cortico-cortical afferents into cortical microcircuits (Hoerder-Suabedissen and Molnar, 2015). Therefore, subcortical structures such as thalamus or basal ganglia (BG) are highly sensitive to perinatal stressors and have been reported to be relevant to neurodevelopment, such as preterm birth (Shang et al., 2019) and autism spectrum disorder (ASD) (Duan et al., 2020; Schuetze et al., 2016). Subcortical brain structures are involved in diverse functions, including motion, consciousness, emotions and learning (Satizabal et al., 2019). Many researchers are aimed to trace the complex developmental processes of subcortices and understand their relationship with functioning deficits in ASD using structural MRI. The surface of subcortices was observed to be positively associated with restricted, repetitive behaviors among ASD participants (van Rooij et al., 2018). ASD children exhibited decreased structural covariation between the left and right thalami (Duan et al., 2020). Another comprehensive mega-analysis demonstrated that ASD participants may have smaller subcortical volumes, which are related to ASD typical symptoms (van Rooij et al., 2018). Accordingly, cortico-striatal-thalamic pathways are considered as common neurobiological substrates across a broad spectrum of neurodevelopmental disorders due to vital components of salience network(SN) (Leisman et al., 2014). Therefore, it is speculated that the damage to the deep nuclei, especially thalamus and BG, may be affected by genes and the environment, resulting in impaired brain development and maturation, and further leading to the occurrence of ASD and related clinical symptoms (Lord et al., 2018).

Although altered subcortical neuroanatomy has been widely documented in individuals with ASD, multiple evidence from brain connectivity suggests that autistic behaviors attribute to deficits in distributed brain networks rather than several regions of the brain. However, a few knowledge is available about the subcortices and their connections with the cortex in ASD across different age groups (Cerliani et al., 2015; Padmanabhan et al., 2013). Furthermore, it is worth noting that there is considerable variation in intelligence quotient (IQ) scores among ASD populations. The majority of affected individuals exhibit intellectual impairment, whereas those on the high-functioning end of the spectrum tend to have average or above average IQ levels.

Misdiagnosis is more likely to occur in high-functioning adults with ASD due to their atypical clinical manifestations and psychiatric comorbidity (Gabrielsen et al., 2018). However, the current understanding of the underlying neural mechanisms in this population is limited. Only a few studies have investigated brain functional connectivity in high-functioning adults with ASD and typically-developing(TD) individuals in a cohort of relatively small sample (Yang et al., 2022; Dong et al., 2023).

Previous intrinsic network studies of ASD have chosen their seeds based on anatomical and functional studies of controls, which may lead to a seed choice more tailored to controls' brains than to ASD patients. This option is subserved by the pattern of stronger connectivity seen in the controls than the ASD group. An alternative approach is to use a data-driven method of identifying each subject's default network, such as independent component analysis (ICA). Different from the purely data-driven ICA approach, seed-based analyses identify correlations in resting brain activity of a particular subcortical seed (such as thalamus and BG) with other brain areas. When ICA identifies subcortical networks affected by ASD and seed-based analysis estimates the specific brain areas associated with the subcortices, both approaches should give supplemental information.

Considering subcortical structures such as thalamus or BG were highly sensitive to perinatal stressors and have been reported to be relevant to ASD, the present study aimed to investigate subcortical networks of thalamus and BG networks in individuals with ASD. Our primary hypothesis was that atypical neurophysiological findings of ASD patients were caused by dysconnectivity between subcortical structures and cortical mantles. To address these issues, we investigated subcortical networks in a cohort of relatively large sample of high-functioning with ASD adults (n = 74) and matched typically developing (TD) subjects (n = 63) using ICA. We then treated the subcortical networks as seeds to perform seed-based FC. We expected abnormalities in the connectivity of thalamus and BG network, since ASD is a neurodevelopmental disorder. Second, we hypothesized that autism severity affected in connectivity strength in the subcortices within the ASD group. To further investigate the relationship between the intrinsic functional connectivity (iFC) of each subcortical network and the severity of ASD, correlations were calculated between scores on the full-scale intelligence quotient (IQ), Autism Diagnostic Observation Schedule (ADOS) and social responsiveness scale (SRS). Finally, we investigated if the subcortical connectivity and clinical features were available to predict ASD using multivariate classification techniques.