Evaluation of changes in the propagation of electrohysterographic signals of uterine records for predicting preterm birth

Letno se prezgodaj rodi veliko dojenčkov pred dopolnjenim 37. tednom nosečnosti. Analiza EHG signalov je že pokazala obetajoče rezultate pri napovedi prezgodnjega poroda dovolj zgodaj v nosečnosti, da bi lahko prezgodnji porod preprečili. V tem magistrskem delu smo razlike med prezgodnjimi in terminskimi materničnimi posnetki iskali v spremembah intenzitete in smeri propagacije (razširjanja) EHG signalov vzdolž maternice v horizontalni in vertikalni smeri, ki smo jih ocenili s pomočjo kratkočasovne križne korelacije. Na frekvenčnem območju 0,08⠒5,0 Hz smo hitrosti propagacije znotraj praznih intervalov prezgodnjih (terminskih) posnetkov ocenili na 10,56 cm/s (7,27 cm/s) in na 12,66 cm/s (10,84 cm/s) znotraj intervalov s popadki. Ocenjene hitrosti propagacije se nekoliko razlikujejo od tistih iz literature. Z našo metodo smo s klasifikatorjem QDA uspeli pravilno klasificirati (“Prezgodnji porod” ali “Terminski porod”) 83,96 % praznih intervalov. Pomembni rezultati magistrskega dela so ugotovitve, da so intervali izven popadkov (prazni intervali) zelo pomembni pri napovedovanju prezgodnjega poroda, da je količina merjene materine srčne frekvence pomemben dejavnik pri napovedovanju prezgodnjega poroda in da je možno učinkovito napovedati prezgodnji porod že okrog 23. tedna nosečnosti.Yearly there are a lot of preterm births, i.e. births before the 37th week of gestation is completed. Analysis of EHG signals has already shown promising results in predicting preterm birth in early pregnancy, so that preterm birth could be prevented. In this thesis we have searched for the differences between preterm and term uterine records in the changes in intensity and direction of EHG propagation along the uterus in horizontal and vertical direction, which has been estimated with short-time cross-correlation. On the frequency band 0,08⠒5,0 Hz we have estimated propagation velocity inside dummy intervals of preterm (term) records to be equal to 10,56 cm/s (7,27 cm/s) and equal to 12,66 cm/s (10,84 cm/s) inside intervals with contractions. Estimated propagation velocities differ slightly from the ones in the literature. With our method and QDA classifier we have correctly classified (“Preterm birth” or “Term birth”) 83,96 % dummy intervals. Important findings in this thesis are that the intervals outside of contractions (dummy intervals) are very important when predicting preterm birth, that the amount of measured frequency of maternal heart is shown to play an important role in the prediction of preterm birth and that it is possible to effectively predict preterm birth around the 23th week of gestation