Спектрална, биотичка и фрактална анализа ЕЕГ и ЕКГ сигнала у експерименталним моделима инфаркта миокарда и епилепсије

[M. Vorkapić]

93 str.

Medicine - Physiological Sciences / Medicina - Fiziološke nauke

Acute myocardial infarction (AMI) and epilepsy are responsible for significant morbidity and mortality. Besides clinical significance, another thread connecting AMI and epilepsy is the importance of signal analysis in establishing diagnosis as well as assessing risk for disease progression. The objectives of this study were to investigate spectral, fractal and biotic characteristics of the EEG and ECG signals in animal models of AMI and epilepsy. We used the rat model of isoprenaline-induced AMI and registered concomitantly ECG and EEG signals in baseline conditions and up to 24h after isoprenaline administration. Histological myocardium evaluation was done at the end of registration period to assess AMI development morphology. The spectral, fractal and biotic characteristics of the EEG and ECG were analyzed. Lindane-induced generalized seizures in rats were used herein as experimental model of epilepsy. EEG and ECG signals were recorded simultaneously before and after lindane administration. Ictal EEG parameters were detected and analyzed. EEG, as well as ECG features were extracted using spectral, fractal and biotic analyses and neural networks were constructed in order to test the possibility of automatic detection of seizures. Development of AMI has been confirmed by presence of ST elevation in ECG and histological findings of myocardial necrosis. Significant alterations were determined in EEG spectral power in the alpha, beta and theta bands upon isoprenaline administration. These alterations were significantly correlated with ST elevation. We found significant changes in biotic, fractal and time domain analyses of ECG before and after isoprenaline administration. All perceived changes returned to baseline averages after 24h denoting acute effects. In the epilepsy model, we saw significant changes in spectral EEG power and band power dominance in ictal vs baseline periods. EEG and ECG features based on spectral, fractal and biotic analyses were extracted and neural networks for detection of ictal periods were trained. Neural network based on our features for detection of ictal events showed favorable performances.

Akutni infarkt miokarda (AMI) i epilepsija su bolesti odgovorne za značajan morbiditet i mortalitet u populaciji. Pored njihovog klničkog značaja, još jedna važna veza AMI i epilepsije ogleda se u značaju analize signala za uspostavljanje dijagnoze, procenu rizika i praćenje progresije bolesti. Ciljevi ove studije bili su da ispitaju spektralne, fraktalne i biotičke karakteristike EKG i EEG signala u animalnom modelu AMI i epilepsije. Koristili smo model Isoprenalinom indukovanog AMI kod pacova i paralelno registrovali EKG i EEG signale pre administracije kao i do 24h nakon administracije Isoprenalina. Radi procene razvoja AMI učinjena je histološka evaluacija miokarda na kraju perioda registrovanja signala. Analizirane su spektralne, fraktalne i biotičke karakteristike EEG i EKG signala. Za modelovanje epileptičkih napada korišćen je Lindanski model kod pacova. EEG i EKG signali su registrovani pre i nakon administracije Lindana. Iktalni EEG parametri su detektovani i analizirani. Koristeći spektralne, fraktalne i biotičke analize, izvučene su EEG i EKG karakteristike koje su nakon toga korišćene za treniranje neuralne mreže u svrhu automatske detekcije i predikcije pojave epileptičkih napada. Razvoj AMI potvrđen je pojavom ST elevacije i karakterističnim histološkim nalazom nakon nekroze miokarda. Nakon administracije isoprenalina registrovane su značajne promene u EEG spektru alfa, beta i gama frekventnih opsega. Navedene promene su značajno korelisale sa ST elevacijom. Takođe, našli smo značajne promene u biotičkim i fraktalnim parametrima i paramterima vremenskog domena EKG signala pre i nakon administracije Isoprenalina. Sve promene su se povukle nazad na osnovne vrednosti nakon perioda od 24h što ukazuje na akutne efekte. U modelu epilepsije registrovane su značajne promene u snazi EEG spektra kao i u dominaciji snage pojedinih frekventnih opsega u iktalnoj u odnosu na osnovne vrednosti. Nakon izolovanja EEG i EKG karakteristika, trenirali smo neuralne mreže koje su potom dale zadovoljavajuće rezultate u automatskoj detekciji i predikciji iktalnih fenomena.

EEG, ECG, HRV, nonlinear, Epilepsy, AMI, signal analysis, Bios, Isoprenaline, Lindane

EEG, ECG, HRV, nelinearne metode, Epilepsija, AMI, analiza signala, Bios, Izoprenalin, Lindan

612.1:612.8(043.3)

Tekst