Mahmut Ozer

Short Communication: EEG signals classification using the K-means clustering and a multilayer perceptron neural network model

We introduced a multilayer perceptron neural network (MLPNN) based classification model as a diagnostic decision support mechanism in the epilepsy treatment. EEG signals were decomposed into frequency sub-bands using discrete wavelet transform (DWT). The wavelet coefficients were clustered using the K-means algorithm for each frequency sub-band. The probability distributions were computed according to distribution of wavelet coefficients to the clusters, and then used as inputs to the MLPNN model. We conducted five different experiments to evaluate the performance of the proposed model in the classifications of different mixtures of healthy segments, epileptic seizure free segments and epileptic seizure segments. We showed that the proposed model resulted in satisfactory classification accuracy rates.

Controlling the spontaneous spiking regularity via channel blocking on Newman-Watts networks of Hodgkin-Huxley neurons

We investigate the regularity of spontaneous spiking activity on Newman-Watts small-world networks consisting of biophysically realistic Hodgkin-Huxley neurons with a tunable intensity of intrinsic noise and fraction of blocked voltage-gated sodium and potassium ion channels embedded in neuronal membranes. We show that there exists an optimal fraction of shortcut links between physically distant neurons, as well as an optimal intensity of intrinsic noise, which warrant an optimally ordered spontaneous spiking activity. This doubly coherence resonance-like phenomenon depends significantly on, and can be controlled via, the fraction of closed sodium and potassium ion channels, whereby the impacts can be understood via the analysis of the firing rate function as well as the deterministic system dynamics. Potential biological implications of our findings for information propagation across neural networks are also discussed.

Weak signal propagation through noisy feedforward neuronal networks.

We determine under which conditions the propagation of weak periodic signals through a feedforward Hodgkin–Huxley neuronal network is optimal. We find that successive neuronal layers are able to amplify weak signals introduced to the neurons forming the first layer only above a certain intensity of intrinsic noise. Furthermore, we show that as low as 4% of all possible interlayer links are sufficient for an optimal propagation of weak signals to great depths of the feedforward neuronal network, provided the signal frequency and the intensity of intrinsic noise are appropriately adjusted.

Autapse-induced multiple coherence resonance in single neurons and neuronal networks

We study the effects of electrical and chemical autapse on the temporal coherence or firing regularity of single stochastic Hodgkin-Huxley neurons and scale-free neuronal networks. Also, we study the effects of chemical autapse on the occurrence of spatial synchronization in scale-free neuronal networks. Irrespective of the type of autapse, we observe autaptic time delay induced multiple coherence resonance for appropriately tuned autaptic conductance levels in single neurons. More precisely, we show that in the presence of an electrical autapse, there is an optimal intensity of channel noise inducing the multiple coherence resonance, whereas in the presence of chemical autapse the occurrence of multiple coherence resonance is less sensitive to the channel noise intensity. At the network level, we find autaptic time delay induced multiple coherence resonance and synchronization transitions, occurring at approximately the same delay lengths. We show that these two phenomena can arise only at a specific range of the coupling strength, and that they can be observed independently of the average degree of the network.

Can scale-freeness offset delayed signal detection in neuronal networks?

First-spike latency following stimulus onset is of significant physiological relevance. Neurons transmit information about their inputs by transforming them into spike trains, and the timing of these spike trains is in turn crucial for effectively encoding that information. Random processes and uncertainty that underly neuronal dynamics have been shown to prolong the time towards the first response in a phenomenon dubbed noise-delayed decay. Here we study whether Hodgkin-Huxley neurons with a tunable intensity of intrinsic noise might have shorter response times to external stimuli just above threshold if placed on a scale-free network. We show that the heterogeneity of the interaction network may indeed eradicate slow responsiveness, but only if the coupling between individual neurons is sufficiently strong. Increasing the average degree also favors a fast response, but it is less effective than increasing the coupling strength. We also show that noise-delayed decay can be offset further by adjusting the frequency of the external signal, as well as by blocking a fraction of voltage-gated sodium or potassium ion channels. For certain conditions, we observe a double peak in the response time depending on the intensity of intrinsic noise, indicating competition between local and global effects on the neuronal dynamics.

Investigating the performance improvement of HRV Indices in CHF using feature selection methods based on backward elimination and statistical significance

In this study, the best combination of short-term heart rate variability (HRV) measures was investigated to distinguish 29 patients with congestive heart failure from 54 healthy subjects in the control group. In the analysis performed, wavelet packet transform based frequency-domain measures and several non-linear parameters were used in addition to standard HRV measures. The backward elimination and unpaired statistical analysis methods were used to select the best one among all possible combinations of these measures. Five distinct typical classifiers with different parameters were evaluated in discriminating these two groups using the leave-one-out cross validation method. Each algorithm was tested 30 times to determine the repeatability of the results. The results imply that the backward elimination method gives better performance when compared to the statistical significance method in the feature selection stage. The best performance (82.75%, 96.29%, and 91.56% for the sensitivity, specificity, and accuracy) was obtained by using the SVM classifier with 27 selected features including non-linear and wavelet-based measures.

Epileptic seizure detection using probability distribution based on equal frequency discretization.

In this study, we offered a new feature extraction approach called probability distribution based on equal frequency discretization (EFD) to be used in the detection of epileptic seizure from electroencephalogram (EEG) signals. Here, after EEG signals were discretized by using EFD method, the probability densities of the signals were computed according to the number of data points in each interval. Two different probability density functions were defined by means of the polynomial curve fitting for the subjects without epileptic seizure and the subjects with epileptic seizure, and then when using the mean square error criterion for these two functions, the success of epileptic seizure detection was 96.72%. In addition, when the probability densities of EEG segments were used as the inputs of a multilayer perceptron neural network (MLPNN) model, the success of epileptic seizure detection was 99.23%. This results show that non-linear classifiers can easily detect the epileptic seizures from EEG signals by means of probability distribution based on EFD.

Design and implementation of a domestic solar-wind hybrid energy system

In parallel to developing technology, demand for more energy makes us seek new energy sources. The most important application field of this search is renewable energy resources. Wind and solar energy have being popular ones owing to abundant, ease of availability and convertibility to the electric energy. This work covers realization of a hybrid renewable energy system for a domestic application, which runs under a microcontroller to utilize the solar and wind power. This project is implemented in accordance with available line-electricity. Batteries in the system are charged by either wind power via a small alternator or solar power via an MPPT module. System control relies mainly on microcontroller. Power resources and loads in the system are monitored and controlled in real time.

A new methodology to define the equilibrium value function in the kinetics of (in)activation gates

&NA; A voltage‐gated ion channel is fundamental in generation and propagation of electrical signals in the excitable membranes. Dynamics of (in)activation gates of the ion channel is modeled by first‐order kinetics. The equilibrium value function is crucial in the kinetics of the (in)activation gates for fitting experimental data. We present a new methodology to define the equilibrium value function based on the lowest approximation of the cluster variation method and the static properties in the molecular field approximation. The methodology allows for exploration of the gating dynamics.

Double inverse stochastic resonance with dynamic synapses

We investigate the behavior of a model neuron that receives a biophysically realistic noisy postsynaptic current based on uncorrelated spiking activity from a large number of afferents. We show that, with static synapses, such noise can give rise to inverse stochastic resonance (ISR) as a function of the presynaptic firing rate. We compare this to the case with dynamic synapses that feature short-term synaptic plasticity and show that the interval of presynaptic firing rate over which ISR exists can be extended or diminished. We consider both short-term depression and facilitation. Interestingly, we find that a double inverse stochastic resonance (DISR), with two distinct wells centered at different presynaptic firing rates, can appear.