| Objective:White blood cell classification is an important indicator of clinical testing and an important basis for diagnosing diseases.Most of the laboratories and hospitals use automatic blood analyzers,which have the advantages of accurate counting,can detect more samples and test indicators at one time.However,such equipment also has many limitations in battlefield rescue,earthquake relief,primary medical treatment and other complex environments.(1)The device is bulky and heavy in weight,resulting in poor carrying capacity;(2)The internal structure of the equipment is complex and the vibration resistance is poor;(3)The equipment adopts wet chemical technology and requires a variety of liquid reagent consumables to complete the inspection work.Reagents are not easy to store and liquid reagents are susceptible to freezing failure in low temperature environments.(4)The consumables used are generally used for hundreds of people.After unpacking,they must be used up in the short term.Otherwise,they will expire and will not be used for routine blood tests in a small number of patients.(5)The maintenance of the equipment is expensive and does not apply to the grassroots or remote areas.In order to solve those problems,this paper studies a white blood cell classification system suitable for POCT equipment.The system has the advantages of simple operation,good environmental adaptability,no need for special maintenance and quick output of test results.It can be used not only for rapid detection of hospital bedside samples,routine clinical tests for primary medical care,but also for medical and health systems where battlefields or disaster sites are imperfect,and where there are a large number of patients who need to be diagnosed quickly.It can make up for the shortcomings of traditional blood cell analyzers and lay the foundation for further development of POCT white blood cell sorting equipment.Research content and research methods:In this paper,the fresh human blood samples were set as the target detection object,the microfluidic chip technology,multi-mode imaging optical detection platform and BP neural network algorithm were combined to realize the engineering implementation of the white blood cell classification and detection system.Finally,referring to the results of traditional blood analyzers,the key indicators such as accuracy and precision of white blood cell count and classification detection were evaluated.The research content includes the following four aspects:1.Blood sample processing and microfluidic chip design.This paper used the microfluidic chip as the carrier to design the white blood cell classification system,which eliminates the sample focusing process in traditional flow cytometry methods.First,the blood needs to be hemolyzed and stained.The purpose of hemolysis is to remove the interference of red blood cells on the white blood cells,and not to destroy the overall structure of the white blood cells.The purpose of the staining is to achieve fluorescence image capture of white blood cells.At the same time,it is also necessary to consider factors such as the reaction time should not be too long and the effect is stable,so as to ensure that the system detection is fast and accurate.The design of the microcavity in the microfluidic chip needs to consider the effect of the cell image taken by the camera under the illumination of the light source.If the microcavity is too thick,it will cause serious cell adhesion.If the microcavity is too thin,the number of counted cells is small and the statistical significance is lost.Therefore,the thickness and size of the microcavity must be fully demonstrated to meet imaging and statistical requirements.2.Optical imaging system design.A multi-modal optical path imaging optical system is designed for sample characteristics to achieve single-particle side scattered light(SSC),forward scattered light(FSC),and fluorescence(FL)imaging functions,eliminating the delicate fluid path structure,high sensitivity detectors and complex photodetection systems in traditional flow cytometry.Firstly,the optical path structure of multi-mode imaging design and the choice of source wavelength are introduced;Secondly,it introduces how to set the light-shielding film in the FSC imaging light path,and prevent the FSC light source from directly entering the camera,while allowing most of the cells to scatter light;Finally,the microsphere simulation experiment is used to verify that the forward scattered light can effectively distinguish the size difference of tiny cells,and the SSC and FL imaging optical paths are briefly introduced.3.White blood cell classification image processing algorithm.The white blood cell segmentation and eigenvalue extraction were completed by image algorithm.The database was constructed based on the extracted cell feature values and the artificial cell type results.The BP neural network was used as the classifier to classify and identify the blood sample images.Firstly,the cell image segmentation uses the fluorescence image as the main image processing object,uses the boundary tracking algorithm to find the cell contour and position,and then uses the position information to locate and extract the cell information in the forward scatter image and the side scatter image,and utilizes the boundary for the adhesion cell.The stripping algorithm separates them and records cellular information.Secondly,an effective data set is established and used to set up the BP neural network.4.System performance evaluation research:With the hospital Sysmex XE-5000blood analyzer as a reference,the white blood cell classification system is evaluated for accuracy and precision,and then gradually optimizing the design of each unit module of the system.Based on the above four aspects,this system proposes a white blood cell classification system based on microfluidic and multimodal optical imaging system for the specific detection project of white blood cell classification of fresh human blood,which realizes the key technological breakthrough of the system,which is the battlefield.Rapid on-the-spot detection of white blood cells such as disaster scenes and bedside testing provides support.Result:(1)The optimal hemolysis and staining ratio is 100ml of fresh blood diluted with 10?l of acridine orange(AO)and 10?l of NP-40,and the blood pretreatment effect was the best.At the same time,through calculation and simulation,it is finally determined that the microcavity chip has a microcavity thickness of 100?m and an area of 2 mm×2 mm.The captured images were statistically significant in maintaining the number of cells in the field of view without causing cell adhesion.(2)Considering the characteristics of AO-stained cells,a center wavelength of 480nm is selected as the fluorescent light source.In order to reduce the influence of hemoglobin in the blood on the light transmittance of the sample,an excitation light source having a central wavelength?of 650 nm is selected as the final scattered light source.In the FSC optical path,the combination of the visor and the lens effectively prevents the forward scattered light source from entering the camera while absorbing light from more than 90%of the cells.Since the SSC and the FL illuminate the sample at a certain angle,there is no possibility of directly entering the camera,but a filter is added to the fluorescent light path,which can effectively reduce the scattered light interference emitted by the cell itself under the fluorescent light source.(3)The accuracy of lymphocyte,neutrophil and eosinophils in the BP neural network training test results were above 90%,which is higher than the similar pattern recognition algorithm.The monocyte results are slightly worse,and further improvement is still needed for the treatment of monocytes.(4)Passing-Bablok regression model was used to analysis,100 samples for total white blood cell detection.The results showed that the proportional differences of our system with the results of Sysmex XE-5000 blood analyzer was 0.989,P>0.05 showed that the two groups had consistency in measurement.The CV was less than 4.0%in the repeatability test result of the WBC concentration of 3.0×10~9/L or more,and the value of curve fitting R~2 was equal to 0.9996.The main technical indicators of this system for WBC counting all meet the requirements for clinical testing.In the evaluation of white blood cell counts,a total of 40 blood samples were selected for systematic classification test.Compared with the Sysmex XE-5000 device test results,the proportional differences in lymphocyte,neutrophil and eosinophils were 0.949,0.953,1.077,respectively,and the systemic differences were-0.312,0.648,0.002,respectively.P>0.05 showed that the two groups had consistency in measurement.The results of the repetitive experiments showed that the CV values of eosinophils and monocytes were slightly larger because eosinophils were mainly due to a small proportion in leukocytes,and slight differences in counts could cause large CV values.Monocytes are because the eigenvalue information is not significantly different from other cell eigenvalue information.The other two types of cells had better CV values,all of which were less than 6.5%.Conclusion:The research results of the subject provide an effective method for POCT,realize the key technical breakthrough of the special platform for white blood cell detection,propose the performance evaluation method of white blood cell classification detection system,and expand the application field of microfluidic chip technology.It provides a method and practice reference for the development of POCT white blood cell classification and detection equipment. |
PDF Full Text Request