| Objective：In-field rapid analysis of complete blood plays a significant role on judgingconditions, performing first aid and formulating medical treatment in multipleenvironments of surrounding, battlefield, earthquake zone and so on. Aiming atovercoming deficiencies of analysis speed, portability, environmental adaptation andsupporting convenience, we conducted various researches and proposed the novel designof dry-type complete blood analysis system. Based on density-grad centrifugation andAO-blood cell specific binding technologies, we built the analysis system combined withcentrifugal model of blood, algorithm model, sensor technology, optical design,intelligent signal extraction and processing. Further, comprehensive evaluation ofrepeatability and linearity were conducted for further clinical application.Methods&Contents：Based on comprehensive analysis of demands and functionalparameters of the dry-type analysis system, we applied methods of key parametersimulation, matching of core device and feedback mechanism, experiments ofsingle/multiple factor(s) with multilevel orthogonal and error analysis of optimizationand upgrading of each submodule:I. Design and optimization of signal collection system: First, imaging simulation ofthe light sources was conducted. Central wavelength of the best result was determinedaccording to relationship between layer’s light density distribution and centralwavelength. Second, in order to improve the performance of anti-shake and adaptability,we introduced the method of static scaling imaging, during which scaling and resolutionof the signal collection module were determined based on length ratio between capillaryand the selected CCD. Then, lens was designed and simulated. Finally, relationshipbetween position of excitation wavelength and efficiency of fluorescence was exploredand the best position condition of SNR was found out.II. Design and optimization of the mechanical module: On the foundation of earlyresearches, upgrading and further optimization of sample-in-out submodule, rotationmeasurement submodule and additional feedback submodule were conducted. Weabandoned the dual-motor scheme, and added circular grating to our system to controlthe feedback. Then, adjusting the mechanism for connection and analyzing the errors ofalignment. Finally, we calculated the maximum errors and its influence in the mostcommon condition.III. Design and implementation of the signal processing system: In the paper,extracting features of various categories was based on feature differences of the collectedimages. Then, we built models of edge detection for linear CCD: model of edge detectionand model of centroid calculation for curves. First, image filtering was conducted according to their features. Second, processing images with corresponding methodsbased on whether the transmitted features are inherent. As to processing of thefluorescence, we extracted layer features adaptively after classification based on theinterference area of the image. Finally we compared the results according to differentcolor components and screened them based on statistic and normal range of the medicalparameters.IV. Experiments of researches on variation law of the fluorescence and performancesof the system: First, experiment of single factor and multilevel was introduced toconclude the variation law of the fluorescence; Second, establishing the experimentalplatform and fabricating the prototype, then conducting experiments to evaluateperformance of repeatability and linearity of our system. The former includedmeasurement repeatability of replicated tubes and single tube per sample. Finally,conducting evaluation based on the experimental results of multi-factors.Results：I. Detecting resolution of the signal collection module has reached130cyc/mm,which means ideal precision is7μm. Analyzing time of our system is no more than1minute, and performance of anti-shake has been proved better compared with apparatusin the market.II. Correlation coefficients of measurement results between our system and wet-typeapparatus have been more than0.95, which means the great precision. Variationcoefficients of single tube per sample is no more than3%,which is relatively better thanthat of apparatus in the market. Variation coefficients of replicated tubes per sample areno more than7%, which is approximately the same to that of apparatus in the market. Inconclusion, repeatability of our system is relatively better.III. Underlying features of one-dimensional signal are adequately mined in ouralgorithm, which processes signals based on the interference positions. Variationcoefficients of layer thicknesses after processing are less than3%, which indicates greatrepeatability of the algorithm. In addition, models built in the algorithm can be applieduniversally to the common one-dimensional signals in the time domain.IV. Prototype has been fabricated and large amount analysis of blood samples hasbeen conducted, results of which show that each parameter has met requirements.Conclusion&Prospect：I. Further optimized design and integration of signal collection module: light sources,sample capillary, lens and color linear CCD in our signal collection module are separatedfrom each other. During establishment of the experimental platform, optical adjustingand eliminating errors can only be conducted with the gradually installation of separation modules, which leads to large freedom of the module on four degrees and increases timecosts. Under the premise of enough machining precision and best SNR in the future, wecan fabricate the modules in separate and design better trimming control structure ofgreat accuracy to improve the convenience for adjusting.II.It is significant for algorithm to lay more emphasize on improving error tolerance,multi-platform porting, improving adaptive analysis performance.III. Further expanding the sample size and diversity: samples to be analyzed willneed to be selected based on age, sex, physical condition, sample size and etc. It requiresmore on range of sample to grantee the better repeatability and linearity. |
PDF Full Text Request