| Point of Care Testing(POCT)plays an important role in modern medical treatment which can effectively reduce diagnostic cost.Among various POCT methods,bisosensors based on centrifugal microfluidic have many unique advantages such as less sample consumption,more accurate valve control and minimizing cross contamination and so on.Therefore,people have made extensive and in-depth research on the application of centrifugal microfluidic in the fields of biochemical analysis and immunoassay recently.However,it still has problems of low plasma separation efficiency,poor liquid transfer efficiency on disc,complex operation and low sensitivity which limit its popularization and application in areas with limited resources.Based on the background above,this dissertation aims to achieve rapid,efficient blood separation and rapid,sensitive immunoassay based on centrifugal microfluidic platform.Firstly,this dissertation proposes the ICCRC structure for the design of centrifugal microfluidic channel to overcome the restriction by hematocrit in traditional structure.Secondly,an improved high-precision manufacturing technology is proposed to realize batch manufacturing of microfluidic chips with high precision and high bonding strength.After realizing successful liquid sequential transfer on chip,thirdly,for the first time,Alpha LISA immunoassay based on centrifugal microfluidic is proposed to overcome the problems of long incubation time,low sensitivity and low stability in traditional centrifugal immunoassay.Fourthly,machine learning is proposed to achieve accurate classification of immune detection signals of low concentration.The specific elaboration of the research is as follows:Firstly,the ICCRC structure for the design of centrifugal microfluidic channel is studied.Since traditional "bifurcation structure" has low plasma separation efficiency with high initial hematocrit and "centrifugal single chamber sedimentation structure" may have cross contamination during the extraction of pure plasma,this dissertation proposes a "interconnected-collecting chambers collinear with rotational center"(ICCRC)structure with plasma reservoir and cell reservoir connected with each other and collinear with the center of the circle,and the volume ratio of them is preset.The physical quantities and physical fields for the structure are modeled for computational fluid dynamics simulation,the simulation results show that ICCRC structure effectively solves the limitation of initial hematocrit on the separation efficiency,and the position of the inlet of the siphon channel is optimized to realize the extraction of purity plasma.These experimental results verify the effectiveness of the structure and provide a solid theoretical basis for rapid and efficient blood separation and subsequent high sensitivity immunoassay based on blood separation.After that,high-precision injection molding combined with ultrasonic bonding for centrifugal microfluidic chip are studied.Since the chip produced by traditional injection molding combined with thermal bonding technology have mucosal phenomenon and low bonding strength,this dissertation proposes high-precision injection molding combined with ultrasonic bonding technology to obtain stable,repeatable and high-precision chips with multi-microchannel structures.Put the precisely manufactured chip in the prototype for blood separation test,the results of high plasma separation efficiency(99%),high plasma recovery rate(32.5%),high plasma purity(99%),low separation time(30 seconds)and low rotational speed(4500 rpm)are obtained.The results verify the effectiveness of the technologies and again verify that the ICCRC structure can overcome the restriction of high hematocrit on separation efficiency in the real blood separation experiment.Then,amplified luminescent proximity homogeneous assay(Alpha LISA)based on centrifugal microfluidic is proposed for the first time.Since the probes used in the traditional centrifugal immunoassay system have low sensitivity and wide spectrum,this dissertation proposes to combine centrifugal microfluidic with Alpha LISA immunoassay for the first time,and Alpha LISA probes with good optical stability,narrow emission spectrum and high sensitivity are successfully prepared.The transmission electron microscope images,emission spectra and Zeta potential diagrams of the probes show that the probes with good dispersity,good optical stability,narrow emission spectra and good biocompatibility have been successfully prepared.The feasibility of quantitative detection of pepsinogen using the probes in commercial instruments is verified,and the results show that the prepared probes have high sensitivity and specificity,which provide a foundation for subsequent “one-step” detection of pepsinogen in whole blood on the Alpha LISA immunoassay enabled centrifugal microfluidic system.Furthermore,aiming at the problem that the traditional immunoassay instruments are bulky,a complete optical system for Alpha LISA-enabled centrifugal microfluidic system is designed and the corresponding “all in one” machine is manufactured.The performance of each key technology in this dissertation is comprehensively verified from linearity,stability,limit of detection,specificity and other aspects.The results achieve a linear range of 1 ng/m L to 150 ng/ m L for pepsin I and a linear range of 1 ng/m L to 100ng/m L for pepsin II with good linearity,low limit of detection,high stability,high specificity and high accuracy for “one-step” whole blood test.The results verify the effectiveness of the whole key technologies in achieving fast,highly efficient blood separation and high sensitivity immunoassays.Finally,aiming at the problem that the system may not be able to accurately distinguish weak positive and negative samples(which may lead to false diagnosis)when detecting samples of low concentration,machine learning is proposed to realize qualitative diagnosis and quantitative analysis of weak positive and negative samples of cardiac troponin I(c Tn I).Firstly,different pre-processing is done according to the characteristics of biological signals;Secondly,different machine learning algorithms are used to train and test the data,and the classification of four clinically significant concentrations(0.02 ng/m L,0.04 ng/m L,0.08 ng/m L and 0.1 ng/m L)is realized.Finally,combining the performance of various algorithms,algorithm cost and clinical requirements for the accuracy of low concentration classification,we choose random forest(accuracy of 92%)to accurately distinguish the weak positive and negative samples of c Tn I.The experimental results show that the machine learning further improves the sensitivity of the qualitative detection and the accuracy of the quantitative detection,expands the clinical application scope of the system,and improves its practicability. |
PDF Full Text Request