The Design And Implementation Of Photoelectric Automatic Feces Analyzer Software

The fully photoelectric automatic feces analyzer is an automated medical testing equipment which integrates optics,mechanics,electronics,software and algorithms.It has important value in the clinical detection of digestive diseases.It mainly combines the physical examination of traits,artificial microscopy detection and biochemical detection into a device to automate the whole detection process,thereby reducing human error and biochemical contamination of feces,besides improving detection efficiency.The design of the fully photoelectric automatic fecal analyzer software mainly includes a communication module,a detection service module,a rights management module,a data processing module,a state maintenance module,and a client module.The whole system is based on the Windows 7 operating system,using the C# language,using the.NET framework,and relying on Visual Studio 2015 for development.The main research work and results of this thesis are as follows:First of all,the communication module of this thesis can be divided into CAN bus communication and TCP communication.The CAN bus communication is the main communication mode.This thesis designs different data frame meanings and writes a timeout retransmission mechanism to ensure stable and efficient transmission of communication modules.In addition,the IO efficiency of the communication module is improved by means of asynchronous IO by means of multi-threading technology.In addition,in this thesis,a new concurrency detection business process is designed for the photoelectric automatic fecal analyzer detection service which is based on the critical path algorithm,multi-thread synchronization technology,and the detection efficiency is improved about 83% in theory.Besides,for the camera acquisition module,this thesis uses the two-level cache to read the image data,which effectively solves the problem of image distortion in the process of image acquisition.Then,because of the characteristics of thick sample layer,the complex image background and the dark brightness in the stool sample,the traditional method of sharpness evaluation can't be used directly.This thesis proposes a new image sharpness evaluation function based on BP neural network which be combined with four image sharpness evaluation methods.According to the Pearson product moment correlation coefficient(PLCC)between the training results and the manual discriminant results,the new definition evaluation function based on the neural network has stronger correlation with the artificial discriminant result,which can be better used for autofocus in the complex microscopic environment of feces.Finally,this thesis presents the system client interface.At the same time,the software system is divided into five states for the system operation and maintenance,which limits the running permission of each state,it can effectively guarantee the stability of software operation.Finally,the system was tested in the hospital field.The test results by 1000 samples show that the system can run stably at high speed.The detecting speed can be 55 samples of an hour,and the detection accuracy can be more than 90%.