Analysis Of Cortical Epileptiform EEG Using Autoregressive Power Spectra

Post traumatic epilepsy(PTE) is a traumatic brain injury of human common complications, but the basic mechanism especially chronic epilepsy after brain injury occurrence, development is still poorly understood at present. Research has shown that, the iron ion of epileptic rat cortical injection induced seizure model may be closer to the clinical manifestations. These animal models are sometimes more than a few months and gradually evolved into a form of epilepsy seizures of various chronic spontaneous EEG and brain wave frequency. In order to progress in understanding the driving seizure basic cellular and molecular mechanisms, but also in order to more accurately assess the process of epileptogenesis, need to design a long-term monitoring system, including EEG monitoring and action of synchronous video monitoring of each attack. One purpose of this study is a three channel radio telemetry system is described, using this system can record the spontaneous EEG seizure activity from the cortical surface in episodes of "peak". This method allows a continuous record before, seizure interictal, and after the onset of the recovery period, allowing the iron ion monitoring induced status epilepticus lasting several months, at the same time synchronous video monitoring, performance of epileptic seizures in rats, and to diagnose diseases according to various forms of.Posttraumatic epilepsy(PTE) is a common consequence of traumatic brain injury(TBI) and significant predictor of poor prognosis in TBI patients. To develop clinical interventions for PTE risk reduction, there is a need to elucidate the epileptogenic mechanisms induced by brain injury. The iron-induced rat model of epilepsy mimics many aspects of human PTE. Intracortical injection of iron results in local neuronal damage and the establishment of an epileptic focus, leading to chronic spontaneous electroencephalographic(EEG) signals and motor seizures, with progressively increasing frequency over many months. Identifying unique aspects of PTE seizure semiology for prognosis and treatment may be aided by novel methods of EEG analysis. In this study, autoregressive(AR) methods were compared to the conventional fast Fourier transform(FFT) for processing EEG signals in iron-induced epilepsy. Power spectra obtained using AR showed higher frequency resolution over a given epoch than the spectra obtained using FFT. Moreover, changes in total AR spectral power and frequency distribution over brief successive periods provided convenient indexes for long-term monitoring of seizures. Autoregression analysis may prove complementary to FFT for EEG analysis in PTE patients.The monitoring process of epileptic seizures is a time-consuming and inefficient work, in reality, most of the time can not predict the onset of epilepsy. This paper introduces a new application of power spectrum of autoregressive(AR) prediction method of spectrum technique scalp EEG seizures. Autoregressive AR model using Marple algorithm to establish normal and epileptic EEG based on Matlab platform; order estimation and the various parameters of the AR model; application model parameters of spectrum estimation for two kinds of EEG(AR spectrum). Measuring the normal cognitive brain regions(frontal and occipital) EEG, establish the data file; EEG for the temporal lobe epilepsy patients in the same areas of the brain to create data file; the above two kinds of EEG preprocessing, noise removal and eliminate artifacts; of the above two categories of EEG based AR model; estimation of the above two kinds of EEG of AR spectrum, and FFT power and their spectra were compared; analysis of normal people and patients with epilepsy in the AR model, the AR spectra of brain cognitive areas of the brain. Normal EEG baseline is relatively stable, smaller amplitude relative to the baseline; epileptic EEG baseline is not stable, relative to the baseline of larger amplitude. Normal EEG for the alpha wave, frequency range is 10-12Hz; epileptic EEG for the Delta and theta, bands in the range 0-5Hz. The AR spectral ratio FFT power spectrum smoothing, AR spectral resolution higher than FFT power. With the traditional Fourier transform(FFT), AR EEG spectrum compared with FFT power, FFT power spectrum is the Fourier transform of each signal point can be determined, and its power spectrum is discrete spectral values, jump up and down violently FFT, variance performance is not good, low resolution, spectral line not smooth. The estimated AR model spectrum smoothing, EEG and AR spectrum, high resolution, variance with improved performance. But the spectral lines of AR similar to the FFT power spectrum or reflect the trend of FFT power spectrum. Experiments show that a AR spectrum performance is better than the FFT power spectrum, on the other hand also verified the correctness of AR model. This study is divided into three parts, each part is as follows. PartⅠ Continuous electroencephalographic monitoring with radiotelemetry in a iron-induced epilepsy rat model to evaluate electrographic seizuresObjective: The combined evaluation of physiology and behaviour allows a complete and more comprehensive preclinical assess ment of central nervous system(CNS) function. An integrated video-telemetric electroencephalography(Video-t EEG) system, which allows the simultaneous and continuous recording of EEG and video images for long periods, was developed. Results:. Discussion: This present study is concern ed with the refinement of the surgical technique, as well as the integration and synchronisation of the commercially available Dataquest telemetry system and the Noldus video system, in order to study cortical, hippocampal EEG waveforms, in combination with behaviour and locomotion. The application of this integrated Video-t EEG system could provide advantages in the ethical use of animals in different pre-clinical research areas.Methods: This study focuses on the refinement of the surgical methodology for the combined recording of cortical, hippocampal EEG waveforms in freely moving rats. The postoperativ e recovery of animals was monitored by recording EEGs by telemetry and the general activity by video, on days 1, 6 and 15 after surgery, for approximately 24 h, on each occasion.Results: The results suggested that the applied surgical technique for the implantation of the telemetric transmitter, allowed for a gradual recovery of the animals within 15 days. During the recovery period the behavioural and locomotor parameters measured, indicated that there were no changes to the light-dark circadian cycle, and these parameters gradually tended to reach background levels within a 15-day period. Using a mechanical connection between the deep and the telemetric electrodes, 15 days after surgical implantation the recording system was able to acquire cortical and hippocampal EEG traces of good quality.Conclusion: This present study is concern ed with the refinement of the surgical technique, as well as the integration and synchronisation of the commercially available Dataquest telemetry system and the Noldus video system, in order to study cortical, hippocampal EEG waveforms, in combination with behaviour and locomotion. The application of this integrated Video-t EEG system could provide advantages in the ethical use of animals in different pre-clinical research areas. PartⅡ Autoregressive Spectral Analysis of Cortical Electroencephalographic Signals in a Rat Model of Posttraumatic EpilepsyObjective: Identifying unique aspects of PTE seizure semiology for prognosis and treatment may be aided by novel methods of EEG analysis. In this study, autoregressive(AR) methods were compared to the conventional fast Fourier transform(FFT) for processing EEG signals in iron-induced epilepsy.Methods: Intracortical injection of iron results in local neuronal damage and the establishment of an epileptic focus, leading to chronic spontaneous electroencephalographic(EEG) signals and motor seizures, with progressively increasing frequency over many months. Aided by the Dataquest A.R.T. Analog software provided by DSI-Transoma Medical(Arden Hills, MN). Through Threhold, Minimum Spike Duration, Maximum Spike Duration judgment epileptic Spike wave, such as Minimum Spike Interval, Maximum Spike Interval, Train the Join Interval judgment such as epilepsy wave cluster. Novel methods of EEG analysis, autoregressive(AR) methods were compared to the conventional fast Fourier transform(FFT) for processing EEG signals in iron-induced epilepsy.Results: Shortly after Fe Cl2 injection(50 ± 18 s, mean ± standard error), epileptiform discharges were observed. The frequency reached 30 Hz, with the wave amplitude exceeding 250 μV. In addition, overt inhibitory waves appeared following seizure episodes. Electroencephalographic seizures began 15 ± 24 min after the first iron injection and lasted for 10.3 ± 0.73 h. Interictal spike activity persisted for the duration of recording(8 weeks). Spontaneous recurrent seizures began within approximately 2 weeks and were present for about 2 months. In control rats, the power of AR-derived frequency spectra was concentrated below 10 Hz(Fig.3A). In PTE model rats(Fig.3B), interictal activity was often observed, with the spectral power of corresponding channels shifted to higher frequencies(> 10 Hz).Conclusions:Our results showed that power spectra obtained using AR had higher frequency resolution over a given epoch compared with those obtained using FFT, for cortical EEG of PTE model rats. Moreover, changes in total AR spectral power and frequency distribution over brief successive periods provided convenient indexes for long-term monitoring of seizures. These findings suggest that autoregression analysis may be used to complement FFT for EEG analysis in PTE patients. PartⅢ The characteristies and clinical significance of the Cortical Electroencephalographic Signals by Autoregressive Spectral AnalysisObjective: Autoregressive AR model using Marple algorithm to establish normal and epileptic EEG based on Matlab platform; order estimation and the various parameters of the AR model; application model parameters of spectrum estimation for two kinds of EEG(AR spectrum).Methods: Measuring the normal cognitive brain regions(frontal and occipital) EEG, establish the data file; EEG for the temporal lobe epilepsy patients in the same areas of the brain to create data file; the above two kinds of EEG preprocessing, noise removal and eliminate artifacts; of the above two categories of EEG based AR model; estimation of the above two kinds of EEG of AR spectrum, and FFT power and their spectra were compared; analysis of normal people and patients with epilepsy in the AR model, the AR spectra of brain cognitive areas of the brain..Results: Normal EEG baseline is relatively stable, smaller amplitude relative to the baseline; epileptic EEG baseline is not stable, relative to the baseline of larger amplitude. Normal EEG for the alpha wave, frequency range is 10-12Hz; epileptic EEG for the Delta and theta, bands in the range 0-5Hz. The AR spectral ratio FFT power spectrum smoothing, AR spectral resolution higher than FFT powerConclusions:Normal human frontal brain electrical baseline is relatively stable, relative to the baseline of smaller amplitude, wave line more intensive; epilepsy frontal brain electrical baseline is not stable, relative to the baseline of larger amplitude, wave line is sparse. Can infer the brain frontal occipital epilepsy incidence area. Normal and epilepsy frontal brain electrical AR spectrum and FFT spectrum are different, the normal forehead pillow power EEG mainly concentrated in the 10-12 Hz, namely the alpha band, only a very small part of the energy is concentrated in 0-5Hz; and epilepsy patients with frontal occipital power EEG main concentrated in the 0-5Hz, namely the Delta and theta bands, only a very small part of the energy is concentrated in 10-12 Hz. Application of final prediction error(FPE), when the order of K increased by 1, FPE(k) will have the minimal value in a K, the K at this time as the order p the most appropriate. AR EEG spectra of FFT power graph and FFT power spectrum is the Fourier transform of each signal point can be determined, and its power spectrum is discrete, we can see from the picture that its spectral value jump up and down violently, variance performance is not good, low resolution, spectral line is not smooth. The estimated AR model AR spectrum lines smooth, high resolution, variance performance improvement. AR spectral line similar to the FFT power spectrum or reflect the trend of FFT power spectrum. On one hand, AR spectral performance is better than the FFT power spectrum, on the other hand also verified the correctness of AR model.