| Objective To develop and evaluate the computerized automatic analysis system of placenta ultrasound image and umbilical Doppler signals. It included three parts:(1)To develop the automatic classification system of placenta image and study its relationship to placenta quantitative structure.(2)Study the umbilical artery and vein blood flow change under acute maternal sheep ischemia; multiple method analyze fetal umbilical artery Doppler signals.(3)Combination use of computerized analysis system of placenta image and umbilical blood flow in clinical work. Material and Methods 1. To develop the automatic classification system of placenta image and research its relationship to placenta quantitative structure. 188 cases of pregnancy women were included. Firstly, traditional Grannum method was used to classify the placenta image, then these images were stored by 576×432dpi and BMP format in computer Pentium 200. 115 cases were regarded as train set and the region of interest are collected. The parameters to distinguish tissue types were identified by four algorithms known as grey scale, frequency two-dimensional filtration and fraction and wavelet transformation. The classifier designs (Fuzzy theory, quadratic, linear and neural network) were chosen to develop the automatic classification system. Finally, the trained classifier was tested to estimate classification performance. Among 188 cases, 26 placenta were collected immediately after delivery. Microscopic structure of vascular-synthetic membrane(VSM), synthetic trophoblast knot, intervillous space, villous areas, fibrin and fetal capillaries were counted respectively and correlated to automatic classifier. 2. Acute maternal sheep ischemic caused umbilical cord blood flow change and multiple method analyze fetal umbilical artery Doppler signals. Two pregnant sheep, twin, 120-125gestation age, weighted 50-55kg were in experiment use. After anesthesia, lamb No.1 was ligated one umbilical artery. Lamb No.2 was ligated cotyledon one by one. Lamb No.3 after cotyledon ligated, the cord was spiraled several rounds to form false knots. Lamb No.4 directedly spiraled the cord without being ligated cotyledon. All blood flow change are recorded and observed. 71(57 normal and 14 abnormal) cases of umbilical artery Dopplersignals were studied by five method(S/D spectrum, pole distribution, fraction and Teager energy and Fisher analysis), sensitivity and specificity are compared.3. Clinical use sixteen normal and twenty abnormal pregnant women (IUGR)took ultrasound examination of placenta images and umbilical cord blood flow every two week. Blood estriol (E3) and human placenta lactin (HPL)were measured at the same time. After delivery, placenta were examined for quantitative structure analysis. All methods were compared.Results 1. As regards degree I and II placenta , no significant statistical difference between four classifier with traditional classifier(P>0.05) was found. As regards degree III placenta, except neural network method, the other three methods showed difference about 1.0 degree with traditional method (P<0.05). The position of placenta had no influence on classification. Quantitative analysis of placenta constituents showed the dynamic positive correlation of automatic classifier with density of villi, number of capillaries, VSM, synthetic trophoblast knot and fibrin. 2. When maternal blood flow was reduced less than 60% of the total flow of placenta, the lamb umbilical artery had transient change with resistance and pulsive rate increase which recovered promptly. Once blood flow decreaseed more than the cutoff value, lamb umbilical vein waveform changed, and lamb was very sensitive to torsion of cord under this ischemic condition. Among 71 pregnant womem, only one fetal with IUGR showed umbilical vein pulsation. Fisher method had the strongest significance in sensitivity (P<0.05) with S/D and other three methods.3. After 30 weeks gestation, blood E3 and HPL had no obvious change while gestati...
|
PDF Full Text Request