Detection of Abnormal Fetuses Using Biorthogonal Wavelet Analysis

Kasim Mohammed, Rand

	Detection of Abnormal Fetuses Using Biorthogonal Wavelet Analysis
ALNAHRAIN JOURNAL FOR ENGINEERING SCIENCES
Article 1, Volume 11, Issue 1, 2008, Pages 153-161
Author
Rand Kasim Mohammed
Abstract
Biomedical signals are generated by complex self-regulating systems that process inputs with a broad range of characteristics. Many physiological time series, such as the fetal heart rate, are extremely inhomogeneous and non-stationary, fluctuating in an irregular and complex manner. In this project, the amplitude of low-frequency fluctuation of fetal heart rate frequency was studied using analysis called wavelet transform. The nature and maturational changes of low-frequency fluctuation of the FHR in normal fetuses were investigated and probability distribution of FHR wavelet coefficients was studied from 28 wk of gestation onward. The value of the parameter a of this distribution did not exceed 1.939 regardless of the gestational age in a normal condition. The value of index a range from 2.1585 up to 3.1652 in fetuses from pregnant women with pregnancy-induced hypertension. This project also presents a system capable of calculating index for the fetal heart rate low-frequency fluctuation distribution value and uses it to identify the fetus condition using the MATLAB 7.3 package. FHR data of 12 normal fetuses and 18 fetuses from pregnant women with pregnancy-induced hypertension all between 28 and 38weeks of gestation were studied. First, the cardiotocography tracing was converted from CTG paper images into digital series using image processing so that the system can analyze it. The FHR was then converted to inter-heart beat time series. Biorthogonal wavelet transform was used to analyze inter-heart beat time series. The histogram of the absolute value of resulting wavelet coefficients was analyzed then probability distribution of wavelet coefficients frequency was used to calculate the fluctuation parameter a. Statistical analysis methods where used to compare between the results. The Kruskal-Wallis test was used to test the significance of the difference among the parameters obtained; Pearson’s test was used to test goodness of fit of the distribution function. In addition, the t-test was used for other statistics. Index a values were used in last stage of the system to identify sick fetuses. When analyzing fetal heart rate of 18 fetuses from pregnant women with pregnancy-induced hypertension the system succeed to identify the presence of a problem in 6 fetuses

Statistics Article View: 92 PDF Download: 44