The study of white matter fiber connections in the brain was mostly accomplished in the past by anatomical staining. However, this method, due to its invasive nature, is only suitable for animal studies and cannot be used to study the living human brain. The rise of diffusion magnetic resonance imaging has made it possible to non-invasively study the white matter fibers of the living human brain without causing damage to the internal tissue structure of the brain. The principle of diffusion MRI is primarily based on the diffusion properties of water molecules. Since water molecules are not free to enter and exit the myelin sheath of myelinated fibers, the form of diffusion of water molecules in myelinated fibers exhibits a high degree of anisotropy. Using this technique, the magnitude of anisotropy within a given voxel can be measured, which can indirectly reflect the degree of myelination or the integrity of the fiber bundle; based on the direction of anisotropy, the direction of the fiber bundle can be traced. Fiber tracking techniques based on diffusion magnetic resonance imaging have been widely used in studies of normal populations and patients with neuropsychiatric disorders to observe changes in white matter fiber bundles in a noninvasive manner. However, the use of diffusion magnetic resonance imaging and fiber-tracking methods to study the organizational patterns of structural connectivity networks in the human brain is still only in its preliminary stage. In 2007, Hagmann et al. used diffusion magnetic resonance imaging to analyze the structural data of two subjects and established an individual-based brain structural connectivity network with about 1000 nodes, and demonstrated that the network has a “small-world” character and its node degree distribution follows a power law distribution. Subsequently, Ituria-Medina et al. used diffusionweighted MRI to construct a weighted structural network of the human brain in 20 subjects. In this study, they used an a priori brain atlas to divide the brain into 90 brain regions, and then measured the probability of connectivity between any two regions to construct a structural network of the brain. The network analysis revealed that the brain network has a “small world” nature, and the node degree distribution of brain regions follows an exponentially truncated power law distribution. They further found that the core nodes in the network were mainly distributed in the amygdala, precuneus, insula, superior parietal lobe and superior frontal gyrus. In 2008, Gong et al. collected a large sample of diffusion tensor magnetic resonance data from 80 subjects, and used the same brain mapping to divide the cerebral cortex of each subject into 78 regions, and established the brain structure network of each subject by setting a reasonable threshold value for the number of fiber connections between the brain regions, and then obtained the average brain structure network of the 80 subjects. By analyzing the network, they found that the network was a “small-world” network, and the node degree distribution obeyed an exponentially truncated power law distribution, which is consistent with the diffusion magnetic resonance imaging-based brain network study by Ituria-Medina et al. In addition, this study found that the core nodes of the network were predominantly distributed in the joint cortical regions of the brain, but the most central regions were in the precuneus and medial frontal regions. Most of the core connections in the network were connected to core nodes distributed between different hemispheres or different brain regions of the same hemisphere, providing a structural basis for the differentiation and integration of brain functions. In the same year, Hagmann et al. used diffusion spectrumimaging (DSI) technique to establish a weighted brain structural network of 5 subjects including 998 and 66 brain regions respectively, and described the core regions in the brain structural network from different perspectives by calculating the node degree, median centrality and node efficiency of the network. Nodal Degree The results of node degree calculation showed that the core nodes of the network were concentrated in the medial parietal lobe, medial frontal lobe and superior temporal gyrus, etc. The analysis of median centrality and node efficiency showed that the centers of information transmission in the network were mainly located in the medial regions of the cerebral cortex, such as the precuneus, the posterior cingulate gyrus, and so on. Further analysis of network modularity revealed that the brain structural network could be divided into 6 modules, and the brain regions (connectors) connecting the different modules were mainly distributed in the anterior cingulate gyrus and precuneus, while the core nodes within the modules were distributed in the frontal, temporal and occipital lobes. Recently, Li et al. used the DTI technique to investigate the relationship between individual human intelligence and the properties of brain structural networks. The researchers first assessed and recorded 79 subjects’ FullScale IQ (FSIQ) based on the Wechsler Adult Intelligence Scale, and then obtained the structural networks of individual subjects based on the deterministic tracing method proposed by Gong et al. By calculating the attributes of the brain structure network and analyzing the partial correlation with the subjects’ FSIQ scores, it was found that the attributes of the brain network were significantly correlated with the FSIQ scores: the higher the IQ scores of the subjects, the more edges the brain network had, the shorter the average shortest path length, and the higher the global efficiency of the network. These results suggest that an individual’s intelligence level is significantly correlated with the topology of his or her brain structure, and subjects with efficient brain structure networks tend to have higher intelligence levels. Yan et al. also used the method proposed by Gong et al. [35] to investigate the relationship between brain structural networks and gender, brain volume and intelligence level, and found that the local efficiency of brain structural networks in female subjects was significantly higher than that of male subjects, and that in females, the local efficiency of brain structural networks and brain volume were significantly negatively correlated with intelligence level. Recently, Gong et al. proposed a probabilistic fiber tracing-based method to construct structural connectivity networks in the human brain, and investigated the effects of age and gender on the topological properties of brain structural networks. They found that the connection density and global efficiency of the brain structural network gradually decreased with age, and the regions with the most significant weakening of efficiency were concentrated in the parietal lobe, while the frontal and temporal lobes showed a tendency of increasing efficiency. At the same time, the researchers also found that women’s brain structural networks have higher connectivity efficiency compared to men’s. The information from the data obtained by diffusion magnetic resonance imaging can be used to non-invasively reconstruct the white matter fibers of an individual human brain, thus clarifying the structural location and walking characteristics of the white matter fibers between brain regions. Therefore, compared with the brain structural network based on morphological indicators, the structural network constructed based on diffusion magnetic resonance data can more intuitively portray the real structural connections between brain regions. However, due to the limitations of magnetic resonance imaging equipment and imaging techniques, there are still many problems in the reconstruction of brain white matter fibers. For example, existing fiber tracing methods (e.g., deterministic tracing methods) still have difficulties in reconstructing crossed fibers and longer fibers, resulting in the loss of some of the connections between brain regions; on the other hand, some probabilistic fiber tracing methods, although they can overcome the above difficulties, will inevitably reconstruct some pseudo-connections that do not exist. Therefore, how to accurately reconstruct the white matter fibers becomes the key to constructing the brain structural network based on diffusion magnetic resonance imaging, which is also one of the core issues in the research of diffusion magnetic resonance imaging technology.