| With the development of society and changes in the environment,the incidence of common blood diseases is increasing rapidly.In 2017 and 2018,the total number of visits to the Chinese Hematology Hospital was 282,000 and 384,900.Statistics show that the incidence of leukemia in China is 2.76 / 100,000,and the 5-year survival rate of patients is only 24.8%.Leukemia ranks first in the incidence and mortality of malignancies in children and under 35 years of age.Blood routine is one of the three major examinations in the laboratory.It mainly uses the blood cell analyzer and microscopy to complete the diagnosis of blood system diseases.However,the blood cell analyzer can only perform cell counting and classification according to its principle,and cannot identify information such as cell morphology and internal structure.If there are abnormal cells,the instrument cannot distinguish between hematological and non-hematological diseases,and even when the cells have structural abnormalities and the number and size do not change much,the diagnosis will be missed.Therefore,the blood cell analyzer can only be used as a preliminary screening method for blood system diseases.Microscopic observation of blood smears is the gold standard for checking blood quality.Microscopic observation can easily and quickly determine whether he has a blood disease or a blood system disease,which has important clinical application value.At present,this method is still based on manual microscopy,with complicated operation steps,which is not suitable for large-scale development,and the results are easily affected by operator experience.This project proposes a fully automatic blood cell reader,which can assist doctors to fully automate the blood smear morphological examination process.The user only needs to place the blood smear to be tested in a special card compartment in the film reader,and the instrument will automatically perform card loading,automatic focusing,and automatic classification.After reviewing the film,the user only needs to review the blood cell morphology results on the supporting software.After the review,the blood smear inspection is completed.The automatic blood cell reader will greatly reduce the workload of laboratory operators,effectively improve the quality of blood smear microscopy,and have a very important promotion value for improving the role of blood smear microscopy in blood system diseases.The research content of this topic is mainly divided into the following three parts:1)Design and implementation of hardware system.According to the design principles and processes in the medical device research and development process,analyze the functional requirements of the reader hardware system,and then develop a system plan.The detailed design of the hardware circuit around the system scheme is divided into three major modules according to the hardware circuit function: 1.The main control chip selection STM32F407,and the minimum system module is built around the main control chip;2.According to the power distribution requirements,detailed power distribution is formulated.Program.Select a medical switching power supply to obtain a DC 24 V voltage input system,and then obtain5 V voltage as a partial chip and peripheral power input through the step-down chip LM2595-ADJ,and obtain;3.3V voltage as the main control chip power input through the LM1117-3.3 step-down chip 3.The design of the peripheral drive circuit,the high-torque traction component uses the TMC260 chip to design the high-current drive board;the microscopy scanning platform uses the TMC2209 chip to design the quiet anti-shake drive motor board;the main control circuit board and each motor drive board are selected CAN communication.2)Software system design and implementation.The subject software part includes the driver of the lower computer and the software development of the upper computer.The driver of the lower computer is designed and implemented with STM32F407 chip as the core resource and Free RTOS operating system as the framework.The logistic regression function model is used to design a multi-view fast focus algorithm and implement the motor drive module.An S-shaped low-power scanning path is designed for the effective staining area of the blood smear.The white blood cell preparation position is obtained by establishing a coordinate system,and then the platform is driven to perform rapid positioning at a high magnification field of view.The software module of the host computer uses QT Creator to design a fully automatic blood cell reader software operation interface,which includes the functions of displaying white blood cell pictures,sample testing,querying results,and data communication.3)System integration and experimental analysis.This part mainly debugs and integrates the above hardware design module and software design module,obtains experimental data through the whole machine experiment,and finally analyzes and obtains the whole machine performance and function evaluation.First,power on the hardware circuit module to test.The positive and negative deviation of the data of the key voltage test points on each circuit board does not exceed 0.1V.Then perform a data communication test between the upper and lower computers.By setting blocking and non-blocking test instances,the packet loss rate is 0%.A motor driver was implanted on a well-functioning circuit system,and 535 blood smear samples were repeatedly tested on the scanning platform.Each sample obtained 50 low-power field-of-view images for white fine detection and recognition,and then used the obtained white blood cell coordinate information to convert the high-power lens to accurately locate and focus the image acquisition.The test results showed that they met the requirements.Finally,the whole machine experiment was performed on the above blood smear samples,and the following data results were finally obtained: a single sample took about 40 seconds to scan under a 10 x microscope,and an average time to take about 3seconds to complete a 100 x microscope.The sample test took about 3 minutes on average,Instrument classification takes significantly less time than manual classification. |
PDF Full Text Request