Automated Evaluation Of Uterine Peristalsis Parameters

Uterine peristalsis is closely related to women’s menstrual cycle,conception and uterine related diseases.Clinically,doctors use cinematic magnetic resonance imaging to observe the uterine peristalsis,which is subjective,time-consuming and laborious.An automatic estimation system of uterine peristalsis parameters is urgently needed,In order to provide accurate basis for diagnosis.The frequency,direction,amplitude and range of uterine peristalsis will be estimated by cine magnetic resonance imaging in median sagittal plane in the upper part of the body.Firstly,the deep learning network nn U-Net is used to segment endometrium,which can achieve good segmentation effect without manually adjusting the network structure or applying different parameter adjustment strategies.In the subsequent calculation process of uterine peristalsis parameters,sampling points were selected based on the endometrium boundary.To facilitate the estimation of uterine peristalsis parameters,the uterine region was intercepted and aligned until the long axis of the uterine cavity was parallel to the horizontal axis.In the calculation of uterine peristalsis frequency,for the movement of endometrial stripping movements,the Time-pixel intensity sequence at each sampling point on the baseline contour was generated,fourier transform is performed on the pixel intensity variation sequence corresponding to each sampling point to obtain the energy sequence,and then all the energy sequences are superimposed to obtain the dominant frequency as the uterine peristalsis frequency;For the wave conduction of the uterine junction zone,the pixel intensity variation sequence of the sampling points on the junction zone is counted,and then the frequency domain integration and analysis are carried out.The peristalsis range was determined by comparing the signal frequency domain characteristics of each sampling point with the peristalsis frequency of the sample uterus.Construct a signal intensity time difference image and use image thining technique,the skeleton of low signal wave band which can reflect the direction of peristalsis was extracted to determine the direction of peristalsis,and the peristalsis amplitude was calculated by counting the motion of endometrial boundary.Finally,the automatic estimation of several parameters of uterine peristalsis frequency,range,direction and amplitude is realized,the different implementation details are compared.The results show that the final scheme can achieve the objective evaluation of uterine peristalsis law.