1 August 2017 in Chaos: An Interdisciplinary Journal of Nonlinear Science
This study investigates the nontrivial effects of autapse on stochastic resonance in a modular neuronal network subjected to bounded noise. The resonance effect of autapse is detected by imposing a self-feedback loop with autaptic strength and autaptic time delay to each constituent neuron. Numerical simulations have demonstrated that bounded noise with the proper level of amplitude can induce stochastic resonance; moreover, the noise induced resonance dynamics can be significantly shaped by the autapse. In detail, for a specific range of autaptic strength, multiple stochastic resonances can be induced when the autaptic time delays are appropriately adjusted. These appropriately adjusted delays are detected to nearly approach integer multiples of the period of the external weak signal when the autaptic strength is very near zero; otherwise, they do not match the period of the external weak signal when the autaptic strength is slightly greater than zero. Surprisingly, in both cases, the differences between arbitrary two adjacent adjusted autaptic delays are always approximately equal to the period of the weak signal. The phenomenon of autaptic delay induced multiple stochastic resonances is further confirmed to be robust against the period of the external weak signal and the intramodule probability of subnetwork. These findings could have important implications for weak signal detection and information propagation in realistic neural systems.
12 March 2020 in PLOS ONE
A decrease in alpha band power is defined as a hallmark of electroencephalogram (EEG) in Alzheimer’s disease (AD). This study devotes to understanding the neuronal correlates of alpha rhythm slowing associated with AD from the view of neurocomputation. Firstly, a modified computational model of thalamo-cortico-thalamic (TCT) circuitry is constructed by incorporating two important biologically plausible ingredients. One is the disinhibition property between different inhibitory interneurons in the cortical module. The other is the full relay function of thalamic relay nucleus (TCR) to the cortical module. Then, by decreasing synaptic connectivity parameters to mimic the neuropathological condition of synapse loss in AD, the correlation between neuronal synaptic behavior and abnormal alpha rhythm is simulated by means of power spectral analysis. The results indicate that these decreases of synaptic activity, i.e., not only the excitatory synaptic connections from TCR to fast inhibitory interneurons Cfte and from excitatory interneurons to pyramidal neurons Cpxe but also the inhibitory synaptic connections from fast inhibitory interneurons to slow inhibitory interneurons Clfi and from inhibitory interneurons to TCR Ctii, can significantly diminish the peak power density over the alpha band of the thalamic output, which implies that there is a slowing of alpha band. Furthermore, the underlying mechanism behind the alpha rhythmic changes is analyzed using nonlinear dynamical technique. The results reveal that decreases of Cfte, Cpxe, Clfi and Ctii can make the thalamic module transfer from a limit cycle mode to a point attractor mode, which may lead to the alpha rhythm slowing in the modified TCT model. We expect this work can be helpful in identifying early biomarkers of AD’s EEG and understanding potential pathogenesis of AD.
1 December 2012 in Chaos: An Interdisciplinary Journal of Nonlinear Science
In the present work, we investigate the nontrivial roles of independent Gaussian noise and time-delayed coupling on the synchronous dynamics and coherence property of Fitz Hugh–Nagumo neurons on small-world networks by numerical simulations. First, it is shown that an intermediate level of noise in the neuronal networks can optimally induce a temporal coherence state when the delay in the coupling is absent. We find that this phenomenon is robust to changes of the coupling strength and the rewiring probability of small-world networks. Then, when appropriately tuned delays with moderate values are included in the coupling, the neurons on the networks can reach higher ordered spatiotemporal patterns which are the most coherent in time and almost synchronized in space. Moreover, the tuned delays are within a range, and the period of the firing activity is delay-dependent which equals nearly to the length of the coupling delay. This result implies that the higher ordered spatiotemporal dynamics induced by intermediate delays could be the result of a locking between the period-1 neuronal spiking activity and the delay. The performance of moderate delays in enhancing the ordered spatiotemporal patterns is also examined to be robust against variations of the network randomness.
8 February 2018 in Nonlinear Dynamics
12 August 2018 in Cognitive Neurodynamics
30 March 2022 in PLOS ONE
Note that identifying Mild Cognitive Impairment (MCI) is crucial to early detection and diagnosis of Alzheimer’s disease (AD). This work explores how classification features and experimental algorithms influence classification performances on the ADNI database. Based on structural Magnetic Resonance Images (sMRI), two features including gray matter (GM) volume and lateralization index (LI) are firstly extracted through hypothesis testing. Afterward, several classifier algorithms including Random Forest (RF), Decision Tree (DT), K-Nearest Neighbor(KNN) and Support Vector Machine (SVM) with RBF kernel, Linear kernel or Polynomial kernel are established to realize binary classification among Normal Control (NC), Early Mild Cognitive Impairment (EMCI), Late Mild Cognitive Impairment (LMCI) and AD groups. The main experimental results are as follows. (1) The classification performance in the feature of LI is poor compared with those in the feature of GM volume or the combined feature of LI and GM volume, i.e., the classification accuracies in the feature of LI are relatively low and unstable for most classifier models and subject groups. (2) Comparing with the classification performances in the feature of GM volume and the combined feature of LI and GM volume, the classification accuracy of NC group versus AD group is relatively stable for different classifier models, moreover, the accuracy of AD group versus NC group is almost the highest, with the most classification accuracy of 98.0909%. (3) For different subject groups, the SVM classifier algorithm with Polynomial kernel and the KNN classifier algorithm show relatively stable and high classification accuracy, while DT classifier algorithm shows relatively unstable and lower classification accuracy. (4) Except the groups of EMCI versus LMCI and NC versus EMCI, the classification accuracies are significantly enhanced by emerging the LI into the original feature of GM volume, with the maximum accuracy increase of 5.6364%. These results indicate that various factors of subject data, feature types and experimental algorithms influence classification performances remarkably, especially the newly introduced feature of LI into the feature of GM volume is helpful to improve classification results in some certain extent.