| Routine blood testing is the most common and effective testing method in clinical medicine.Many of its specific indicators are sensitive indicators,which can fully reflect the pathological changes of the human body.At present,blood routine detection methods mostly use microscopic counting,coulter counting,flow cytometry,and image analysis.The microscopic counting method uses manual microscopy,which results in low detection efficiency and excessive resource consumption.Equipment is used the coulter counting method and flow cytometry of the microscopic examination method improve the detection efficiency and reduce the resource consumption,but the disadvantages such as expensive equipment cost and bulky cannot be ignored.The image analysis method based on computer processing technology adopts the automatic microscopic examination method,greatly reduces the size of the equipment,but its preprocessing process is cumbersome and the algorithm complexity is high,and it is difficult to meet the real-time requirements of actual engineering applications.Based on the above reasons,this subject combines lensless digital holography technology with GPU technology to develop a fast automatic counting analysis system software that is simple to operate,low-cost,and easy to miniaturize.This topic is based on the existing red blood cell count analysis algorithm,and the algorithm is implemented in parallel under the compiling and running environment of VS2013 and CUDA7.5,making full use of the advantages of Qt5.5 to develop a graphical user interface,an automatic count analysis system software is designed and implemented,which integrates image acquisition,image reconstruction,image analysis and report export.In this software,SQLite is used to manage the database of user information,CUDA programming model is used to accelerate the algorithm of auto focusing angular spectrum reconstruction,lognormal distribution curve fitting algorithm,binary image reconstruction algorithm and cell counting algorithm are realized by C + +language,and the relevant physiological parameters are calculated.This article carried out a one-week repetitive stress test on the overall function of the software,and there was no downtime,which fully verified that the software has good stability and stress resistance.The software was run on Windows and Linux platforms,respectively.The ideal counting results can be obtained,verifying that the software has excellent cross-platform performance.At the same time,the software has good operation friendliness and scalability,and provides two modes for different usage scenarios: a one-key counting mode that is convenient for users to operate and step-by-step counting mode that is convenient for users to analyze.This article implements the counting analysis algorithm through the popular GPU programming model at this stage.Compared with the traditional CPU programming model,the execution rate of the software has been greatly improved.The operating response time of each functional module is less than 1.5 seconds.Finally,a comprehensive experimental test was carried out on the blood of healthy adults,and the physiological parameters of red blood cells calculated by the system were compared with the clinical results.It was found that the number of red blood cells RBC showed good consistency,and the difference rate was basically controlled within 10%.The results of hematocrit HCT and average red blood cell volume MCV had little volatility,but due to factors such as limited experimental conditions and manual operations the overall performance is higher than the clinical value.In addition,the results of the standard deviation of the red blood cell volume distribution width and the coefficient of variation of the red blood cell volume distribution width are relatively volatile.Therefore,the counting analysis algorithm has room for further improvement and optimization in the follow-up work. |
PDF Full Text Request