Oocyte Quality Detection Based On Microfluidic

In recent years,the clinical pregnancy rate of in vitro fertilization is about 35.8%.The quality of oocytes and the timing of oocyte fertilization are important conditions for the subsequent maturation of in vitro fertilized cells.The quality of oocytes determines the success of in vitro fertilization and embryo transfer Rate,which in turn affects the pregnancy rate.At present,the commonly used method in clinic is to allow experienced physicians to inspect the quality of oocytes through microscope observation to judge the timing of oocyte fertilization.However,with the increase in the demand for oocyte detection,the burden of doctors' microscope observation has increased significantly,and there is a certain degree of subjectivity in doctors' detection.Therefore,it is of great significance to find a method that can assist the detection and judgment with the help of the computer to improve the detection efficiency of oocyte quality and enhance the accuracy of judgment.At present,most of the existing methods for detecting oocyte quality have the following problems,for example,some damage will be caused to oocytes during the detection process,and there is a lack of detection data for human oocytes and errors in the detection data.The analysis of the deformability of cells can avoid damage to oocytes,so this paper uses microfluidic technology and image analysis methods to analyze the deformability of oocytes.Previous studies have shown that there is a direct positive relationship between the quality of oocytes and their deformability,and deformability can be used as a parameter to characterize the quality of oocytes.In this experiment,the quality of oocytes was evaluated by the deformability analysis of oocytes when they passed through the microchannels in the microfluidic chip.Not only will this method cause no damage to the oocytes,but it is more convincing to use human oocytes for research.At the same time,the quality of the oocytes can determine the better timing of oocyte fertilization.Using Canny operator for digital image processing and deep learning network to evaluate the quality of oocytes,in order to achieve the accuracy of computer detection and the unity of standards.The main experimental work and contributions of this subject are as follows:(1)A new method for detecting oocyte quality based on microfluidic technology is proposed.This method is mainly based on the analysis of the deformability of the oocyte,combined with the Canny operator and deep learning method to evaluate the quality of the deformability of the oocyte and help the doctor to carry out a better timing of fertilization of the oocyte Judgment.Build an experimental platform for microfluidics,use a high-speed camera and microscope to capture oocytes passing through the microchannel of the microfluidic device at high speed,and conduct many experiments with different cells,and compare and analyze the obtained experimental data to obtain oocyte There is a linear relationship between the shape variable and quality.(2)Use Canny operator to extract the edge of the cell when the oocyte passes through the channel of the microfluidic chip to obtain the length and width of the cell.Due to the different deformability of oocytes of different masses passing through the same microfluidic device,different cell lengths will be obtained for this characteristic,and the shape variable of oocytes can be analyzed using Canny operator,and To achieve the purpose of evaluating the quality of oocytes.(3)The faster?rcnn?inception?resnet?V2 network was used to create a data set of oocytes,label and train the pictures,classify according to different oocyte periods,and verify the experimental data to obtain an average accuracy(MAP)of 97.056%.(4)Compare the method of using Canny operator with the method of deep learning.Through the analysis of Canny operator,it is found that there is indeed a linear relationship between the number of days and the length of oocyte and the R2> 90%.The change trend between the two is similar to linear transformation.The average accuracy(MAP)of the deep learning model training results is as high as 97.056%,indicating that the deep learning model can identify cells in different periods well.These two methods can more comprehensively evaluate the quality of oocytes and can distinguish the different periods of oocytes by analyzing the deformation of oocytes,and can judge the timing of oocyte fertilization based on this.