Research On High-throughput Automatic Microscopic Imaging System For Biochemical Detection

In the medical field,the use of automatic microscopy systems for pathological diagnosis instead of traditional microscope diagnosis can greatly reduce the work intensity of medical staff and increase work efficiency.However,the demand for sample processing in pathological research institutions is often very large,and the annual inspection volume of an ordinary pathological institution slice can reach tens of thousands of cases.According to the "Guidelines for the Construction and Management of Pathology" issued by the National Health and Family Planning Commission,1-2 pathologists are required for every 100 beds,and the gap between Chinese pathologists is 9 times that of existing staff.With such scarce medical resources,it is necessary to further increase the productivity of doctors.When using an automated microscopy system for diagnosis,the time it takes for the microscopy system to acquire images accounts for almost 80%of the physician's diagnosis time.Therefore,further improving the scanning speed of the automatic microscopy system is very important to improve the diagnosis efficiency of doctors.In this paper,we study three aspects of autofocus,focal plane modeling,and microscopic image scanning acquisition methods to improve the scanning efficiency of microscopic systems.The specific research contents are as follows:Firstly,research on autofocus algorithm.The autofocus of the microscope system mainly includes the image sharpness evaluation,focus window selection and focal plane search.Aiming at these three processes,this article has designed an optimized sharpness evaluation algorithm combining fast Fourier transform evaluation function(RFT)and SMD2,an adaptive focus window selection algorithm and a golden ratio-based hill-climbing search method to achieve fast and accurate focusing.Secondly,research on focal plane modeling methods.In the actual scanning process,the Z-direction distribution of the focus of the sample is not irregular.By establishing a focal plane model to predict the focal position,the number of autofocus points can be effectively reduced and the scanning efficiency of the microscopic system can be greatly improved.However,traditional focal plane modeling methods often cannot meet the requirements of high-precision scanning due to low accuracy.Therefore,two high-precision modeling methods are designed in this paper,which can greatly improve the accuracy of the model.At the same time,the optimization calculation method of the plane equation is designed to ensure the modeling efficiency of the focal plane model.Finally,research on micro image acquisition method.The traditional image acquisition method of the microscopy system is to acquire images under the stationary state of the platform,which will cause a pause in the scanning process and seriously affect the scanning speed.Therefore,a scanning method for dynamically acquiring images is designed in this paper,which can greatly increase the scanning speed.Aiming at the problem that dynamic image acquisition affects image quality,this paper designs a sinusoidal variable-speed scanning motion speed curve and a parallel scanning strategy,thereby achieving a fast dynamic image acquisition method that can be clearly imaged.