Autism spectrum disorder (ASD) is a neurodevelopmental disease whose core symptoms are atypical social dysfunction and repetitive, restricted behaviors (Baird et al., 2003). With the increasing prevalence of ASD (Maenner et al., 2023) and the maturity of resting-state functional magnetic resonance imaging (rs-fMRI), a large number of researchers have studied the brain activity of ASD patients using fMRI techniques (Paz, 2019). In particular, some researchers have gradually begun to focus on subcortical structures because of their crucial role in social-emotional communication, behavioral regulation, and advanced cognition (Ayub et al., 2021; Di Martino et al., 2011; Kleinhans et al., 2016).

The striatum, a vital node of information pathways in the brain, is part of the structure of the corticostriatal circuitry. In this circuit, the striatum processes abundant information from the cerebral cortex and transmits the processed information back to the cerebral cortex via the thalamus, involving human emotional regulation, motor control, reward processing, and cognitive control (Haber, 2016; Sarter et al., 2021). Researchers have observed striatal dysfunction in ASD, and some of these studies showed that striatal dysfunction was associated with abnormal behaviors in ASD. For instance, Bentea et al. found that when a genetic defect caused abnormal neurotransmission at striatal synapses, mice developed an autism-like phenotype of repetitive behaviors and reduced social interaction with others (Bentea et al., 2021). When studying the post-mortem brains of ASD patients, Kuo et al. found abnormal protein expression in the caudate nucleus (Kuo & Liu, 2020). Janouschek et al. reported that the striatum of ASD patients generally displayed hypoactivation, which was correlated with atypical reward processing (Janouschek et al., 2021). Therefore, striatal alterations in ASD merit further investigation.

By calculating the correlation between blood oxygen level–dependent (BOLD) signals, functional connectivity (FC) analysis can assess the relationship of neural activity between regions; it has been widely used in research on neuropsychiatric disorders. In studies of ASD, researchers have used this approach to explore the functional changes in the striatum. For instance, Di Martino A et al. found that individuals with ASD possessed specific patterns of striatal FC compared with TD controls by using a seed-based FC approach (Di Martino et al., 2011); Delmonte et al. calculated the FC between the frontal lobe and the striatum and reported abnormalities in frontostriatal connectivity in the ASD group (Delmonte et al., 2013); Polk M et al. explored the whole-brain FC of the nucleus accumbens (NAcc) in ASD patients and found aberrant FC between the posterior cingulate cortex and NAcc (Polk & Ikuta, 2022). These studies all demonstrated that the corticostriatal circuitry was impaired in ASD; and illustrated that FC may be a promising approach to reveal this dysfunction. However, the calculation of FC as mentioned above requires averaging the BOLD signals generated in a particular brain region across the entire image acquisition period. Clearly, this static FC (sFC) ignores the dynamic nature of brain activity.

Even in a resting state, the brain neurons continuously integrate and transmit various biological information, which is time-dynamic. The emergence of dynamic functional connectivity (dFC) technologies allows us to explore this dynamic nature of brain activity. By consecutively intercepting time windows over the entire BOLD time series and calculating FC within each time window, the dFC-Sliding Time Window method can capture changes in brain FC over time, providing new perspectives for diagnosing and treating disease (Hutchison et al., 2013; Thompson, 2018). Several studies of ASD have used dFC to explore dysfunction in brain activity. For example, Gao et al. found altered dFC between the amygdala and the bilateral inferior temporal gyrus, which was correlated with the clinical symptoms of ASD (Gao et al., 2022). He et al. found that patients with ASD had lower dFC between the posterior cingulate cortex and the right precentral gyrus than the control group and that dFC alterations were correlated with abnormal social interaction in ASD patients (He et al., 2018). Moreover, studies related to the diagnosis and classification of diseases have shown that dFC has better classification performance than sFC (Yan et al., 2020). Therefore, it is necessary to explore brain activity using dFC.

Most previous studies have explored the intrinsic brain FC patterns of ASD patients in only a single frequency band (0.01-0.08 Hz), which would overlook meaningful information related to frequency-specific spontaneous fluctuations. Researchers have indicated that the activity of neurons in the brain is frequency specific and that the BOLD time signal in different frequency bands reflects specific biological information (Buzsáki & Draguhn, 2004). In a study by Zuo et al., researchers indicated that low-frequency oscillations in the slow-4 (0.027-0.073 Hz) band were mainly related to neuronal activity in the basal ganglia, while low-frequency oscillations in the slow-5 (0.01-0.027 Hz) band were primarily related to activity in the ventromedial prefrontal cortex (Zuo et al., 2010). When studying the amplitude of low-frequency fluctuations (ALFF) in ASD, Mei et al. found that the age-by-group interaction effect could be tested only in the slow-4 band (Mei et al., 2022). Frequency specificity was also detected in other disorders, including schizophrenia (Yu et al., 2014), Alzheimer's disease (Yang et al., 2020), Parkinson's disease (Wang et al., 2020), and depression (L. Wang et al., 2016). These studies all suggest that measuring brain activity in precise frequency bands can provide new insights into the pathogenesis of diseases.

In this study, we investigated the striatal FC of ASD for its vital role in brain activity. Considering the dynamics and frequency specificity of brain activity, we evaluated the striatal FC of ASD patients in the slow-4 and slow-5 frequency bands from both static and dynamic perspectives. To explore the potential of striatal FC in ASD classification, we also constructed support vector machine (SVM) classification models using the aberrant sFC and dFC feature values. We hypothesized that striatal dFC and sFC were altered in ASD, and these alterations were frequency specific. Meanwhile, the altered striatal dFC and sFC could distinguish patients with ASD from TD.