Internet gaming disorder (IGD) is characterized by an uncontrollable urge to engage in gaming, leading to significant impairments in academic, social, and personal domains (APA, 2013). Research suggests that IGD may be associated with deficits in cognitive control, attention, emotional regulation, and audiovisual processing (Dong & Potenza, 2014; Kuss et al., 2018; Niu et al., 2022).
Neuroimaging studies have linked IGD to functional abnormalities in several brain networks regions, including the default mode network (DMN) for self-referential processing (D. Lee et al., 2017), the ventral attention/salience network (VAN) for top-down attentional control (Frank & Sabatinelli, 2012), and the sensorimotor network for sensory-motor integration (Wang et al., 2019; Yan et al., 2021).
Structural neuroimaging studies further suggest that individuals with IGD exhibit gray matter volume (GMV) abnormalities in key regions, including the cingulate cortex (a core hub of the salience network), the prefrontal cortex (central to executive control), and sensorimotor-related areas (Niu et al., 2022). Additionally, reduced cortical thickness has been reported in the orbitofrontal cortex, insula, and precuneus—regions involved in audiovisual processing and sensory-motor integration (Wang et al., 2018). These findings highlight widespread neural alterations in IGD, providing insight into its neurobiological mechanisms.
However, brain regions do not function in isolation but instead form a cohesive, anatomically interconnected network. Measuring structural similarity across multiple morphological features remains a challenge. Accordingly, morphometric similarity network (MSN) portrays brain regions as a vector of macrostructural indices and construct networks based on pairwise correlations between regional feature vectors, offering valuable insights into this issue (Sebenius et al., 2023; Wu et al., 2023; Yao et al., 2024). Although previous research has identified MSN alterations in IGD (Wei et al., 2021), MSN summarize vertex-wise cortical features into single regional values using standardized statistics, leading to information loss (Sebenius et al., 2023). Additionally, vertex-based MSN construction is typically limited to single morphological features, such as cortical area or volume. (Homan et al., 2019; Kong et al., 2015; Leming et al., 2021).
To address these limitations, the morphometric inverse divergence (MIND) method was developed (Sebenius et al., 2023). Unlike traditional MSN, MIND integrates multiple structural features at the vertex level and computes symmetric Kullback-Leibler (KL) divergence between regions. This results in networks with higher reliability, stronger alignment with cortical cytoarchitecture, and greater correlation with tract-tracing measures of axonal connectivity. Furthermore, compared to diffusion-weighted MSN, MIND networks show greater sensitivity to age-related changes. They also exhibit stronger coupling with gene co-expression patterns and higher heritability, especially edges between structurally differentiated areas. These findings suggest that MIND networks are biologically validated and more closely aligned with the brain’s intrinsic biological and genetic architecture (Sebenius et al., 2023).
Notably, studies on first-episode schizophrenia and major depressive disorder have also reported abnormal MIND network alterations (Gong et al., 2025; Yao et al., 2024). However, no studies have yet investigated MIND-based structural similarity alterations in IGD.
Brain network alterations often involve complex, large-scale interactions, forming an integrated and efficient network. Graph theory provides a powerful framework for analyzing the topological organization of brain networks (Stam & Reijneveld, 2007). Altered network topology has been observed in substance addictions such as heroin use (Yuan et al., 2010), alcoholism (Sjoerds et al., 2017), and smoking (Lin et al., 2015), as well as in gambling (Tschernegg et al., 2013). Previous studies have reported that IGD is associated with high global efficiency and low local efficiency, suggesting a shift toward random network topology (Park et al., 2018). Furthermore, disruptions in network topology have been observed in the frontal, occipital, and temporal lobes in IGD (Chen et al., 2020; Wee et al., 2014). However, whether similar topological abnormalities exist in IGD’s MIND network remains unclear.
Here, we address this gap using a large sample of 110 individuals with IGD to enhance statistical power and reliability. Unlike previous studies, we employed recreational gaming users (RGU) as controls. While RGUs spend similar time gaming as IGD individuals, they lack clinical addiction symptoms, such as compulsive cravings or withdrawal. This design isolates core neural abnormalities in IGD, independent of gaming behavior.
To comprehensively explore the differences in the MIND network in IGD, we compared the mean differences across large-scale sub-networks, as well as the mean differences for each region. Additionally, we examined the differences in each individual connection between regions to identify more subtle changes between them (edge analysis). We further explored topological properties, including small-worldness, efficiency, and nodal centrality.
Given the literature, we hypothesize that IGD will exhibit altered inter-regional structural similarity in several large-scale brain networks: specifically, (1) the ventral attention networks, due to their roles in attentional bias and deficient bottom-up control; (2) the default mode network, associated with self-referential processing and craving; and (3) the sensorimotor network and visual networks, due to their involvement in motor planning and execution during gaming. In addition, we expect IGD to show disrupted topological organization in the MIND network, characterized by decreased local efficiency and increased global efficiency, reflecting a shift toward more randomized and less modular network architecture. Altered nodal centrality is also anticipated in regions involved in attention, salience detection, and sensorimotor processing.
Comments (0)