Study On Automatic Measurement Of Angle Of Descend Based On Ultrasonographic Features Of Fetal Head And Pubic Symphysis

Fetal head station is one of the important parameters in the birth process.Paying close attention to the changes in the fetal head station during the birth process will help doctors understand the situation of delivery and make corresponding clinical interventions,so as to predict the delivery mode and reduce maternal and infant injuries.However,the subjective and invasive traditional vaginal digital examination to assess the fetal head station has a large error rate,is not reliable,and may cause maternal infection.In recent years,relevant studies have shown that monitoring labor parameters with ultrasound technology is a more effective method.The evaluation of angle of descent(AOD)through transperineal ultrasound provides a more accurate,objective and repeatable method for detecting the fetal head station during labor.However,at present,manual measurement is commonly used in the measurement of AOD in clinical practice,which requires operators to have certain ultrasound experience,and there is also a certain degree of subjectivity among different operators.In view of the above problems,this article proposes a fully automatic measurement method of AOD,including the recognition method of the standard plane of AOD and the automatic measurement method of AOD.The recognition method of the standard plane of AOD is based on the deep learning algorithm VGG.We have simplified and adjusted the algorithm to the task requirements.For the test of 122 transperineal ultrasound images,the accuracy of the recognition algorithm of standard plane is 94.26%,and the average time is only 6.41±0.07 ms.For the automatic measurement method of AOD,we first use the U-Net model-based segmentation algorithm to segment the pubic symphysis and fetal head region.Then,the segmentation results are processed by connected domain analysis to filter out some segmentation errors.Then,the reference points of measurement of AOD in the pubic symphysis were extracted by the method based on principal component analysis.The experimental results show that the method for extracting Characteristic points of pubic symphysis based on the principal component analysis algorithm is better than the method using ellipse fitting.After the reference points of measurement of AOD extraction is implemented,we use the method based on image grayscale to screen the feature points on the contour of the fetal head to avoid the error of the ellipse fitting result caused by incorrect feature points.We use the method of direct least squares fitting of ellipses to obtain the contour equation of fetal head.Finally,calculate the AOD based on the extracted features.Bland-Altman analysis results show that the automatic measurement method of AOD proposed in this paper has a good consistency with the manual measurement method.The automatic measurement method of AOD proposed in this paper provides a more objective and effective method for quantitatively describing the fetal head station,which is of great significance in reducing the workload of doctors and helping doctors make correct decisions in the labor process.