| ObjectiveAlzheimer’s disease(AD) is a progressive neurodegenerative disorder for cognitive dysfunction as the main feature, which brings heavy burden to family and society. How to identify the early changes of cognitive dysfunction and give the appropriate treatment, is of great significant to delay the onset of dementia. Researchs have shown that AD is associated with abnormal changes of structural and functional brain networks. Graph theory has been used to concisely quantify the properties of brain networks, it provides another method for the early diagnosis and treatment evaluation of AD from a complex networks perspective. In this papers, by using resting-state electroencephalography(EEG) data of AD patients and normal elderly peoples, we investigated functional brain networks of the two groups.Then combining the graph theory we quantitatively analysised the topological and rich-club organization of the networks. This study provides support for functional brain state of AD in quantitative analysis from the level of brain networks.Methods16channels EEG were recorded under the resting, eyes-closed condition in 15 AD patients and 15 control subjects. After preprocessing which includes baseline drift correction and power frequency filtering, the datas were used the method of Short Time Fourier Transform for time-frequency analysis. Then the delta band(0.5~4 Hz), theta band(4~8 Hz) and alpha band(8~13 Hz) of EEGs were extracted. The network can be represented by a graph which consists of a collection of nodes(vertices) and links(edges). The synchronization likelihood was estimated on the EEG data after preprocessing. The purpose of the synchronization likelihood analysis is to measure the full set of all pairwise couplings, which resulting the synchronization likelihood matrices. Considering a threshold T which satisfied the condition of a network, the matrices were converted into binary graphs. Finally, the full band networks, delta network, theta networks and alpha networks of the two groups were constructed in the range of the threshold(0.05≤T≤0.07). Then the clustering coefficient and global efficiency of these networks were calculated and compared between the two groups. The clustering coefficient is a measure of the local interconnectedness and global efficiency measures global network information transmission capacity. According the computed rich-club coefficients Ф(k) across a range of degree k of the network, we investigated the rich-club organization of the large-scale functional brain networks.Results 1. Time-frequency analysisThe brain energy distribution of the AD group concentrated in 0.5~10 Hz, and the energy of delta rhythm was larger than the alpha and theta rhythm. The energy of the normal group mainly distributed in 0.5~12 Hz, involving the delta, theta and alpha rhythm, and the energy of the alpha rhythm was largest.2. Construct the network based on the synchronization likelihood analysisCompared with the normal group, the average synchronization lilelihoods of the AD group at the theta band(0.2105±0.0412) were lower than the values of the control group(0.2469±0.0524)(t test, P<0.05).3. The clustering coefficient and the global efficiency of the graphs(1)At the case which the threshold values is T=0.06 and T=0.07, the clustering coefficients of the networks in AD group at the full band(0.8665 ± 0.0545, T=0.06)( 0.7770±0.0884, T=0.07) were both lower than the control group(0.9227±0.0587, T=0.06)( 0.8905±0.0798, T=0.07)(P<0.05). For the values of T(0.05≤T≤0.07), however, there was no significant difference between the global efficiency of the two groups at the full band.(2)For the range of thresholds(0.05 ≤ T ≤ 0.07), there was no significant difference between the clustering coefficient and the global efficiency of the AD group and the control group in delta network.(3)For the values of T(0.05 ≤ T ≤ 0.07), the clustering coefficient were significantly smaller in the AD group of the theta network(0.8939±0.0564,0.8161±0.0838,0.7783±0.1229)compared with the control group(0.9519±0.0644,0.9053±0.0882,0.8644±0.0942)( P<0.05). The global efficiency in the AD group of the theta network(0.9305±0.0493,0.8873±0.0587,0.8267±0.0732)were smaller than the controls(0.9872±0.0171,0.9697±0.0321,0.9403±0.0593)(P<0.05).(4) For the range of thresholds(0.05≤T≤0.07), the clustering coefficient of the alpha network in the AD group(0.8939 ± 0.0564,0.8161 ± 0.0838,0.7783 ±0.1229)were significantly smaller than the control group(0.9519±0.0644,0.9053±0.0882,0.8644±0.0942)(P<0.05). The global efficiency in the AD group of the theta network(0.9347±0.0376,0.8490±0.0720,0.7648±0.1074)were also smaller than the controls(0.9658±0.0427,0.9229±0.0777,0.8704±0.1009)(P<0.05).4. rich-club coefficients(1) For the values of T(0.05≤T≤0.07), there are the rich-club coefficients of the networks is greater than 1 in both AD and control groups.(2) Within the range of threshold value(0.05 ≤ T ≤ 0.07), the unweighted rich-club coefficients of the AD group in the full band, delta, theta and alpha network were significantly different(P<0.05). There was a significant difference in the delta networks of the control group for the same threshold range(P<0.05). When the threshold values of T=0.06 and T=0.07, the unweighted rich-club coefficients of the AD group in the theta network is greater than the control group(P<0.05).For the threshold range(0.05≤T≤0.07), the weighted rich-club coefficients of the AD group were significantly different at the delta and theta networks, and were only at the delta networks of the control group.ConclusionIn the paper, the functional brain networks of the AD and control groups have been constructed based on the resting state EEG by using the synchronization likelihood analysis. And then we have measured these networks combined with the graph theory.(1) The synchronizability of the AD brain networks were lower than the control group, and the clustering coefficient and global efficiency at the theta band and alpha band in AD group was smaller.The results suggest that there may be a loss of the functional brain connections in AD.(2)There are rich-club organizations of the functional networks in both AD and control group, suggesting that highly connected nodes have a strong tendency to be mutually interconnected. The threshold values have a greater influence on the rich-club organization of the AD group, which suggests that connections between the highly connected nodes may lost in AD functional brain network.Using synchronization likelihood and graph theory based on EEG is a possible method to study the characteristics of functional brain connectivity. And the topological parameters and the rich-club coefficients may be of benefit to quantify the state of the brain in AD. |
PDF Full Text Request