Fluorescence Image Processing Of Digital PCR Based On Spectral Unmixing Technology

In recent years,Digital PCR is a new technique for quantitative analysis of nucleic acid.The results of digital PCR reaction are presented in the form of fluorescent spot images.However,due to the influence of many factors such as environment,instrument and samples,there will be some fluorescence interference in the system,which will be superimposed on the fluorescence result image,increasing the difficulty of subsequent image processing and analysis,and affecting the accuracy of the PCR result interpretation.The main interference comes from different non-samples fluorescence spectra in the system,such as background fluorescence.The accurate analysis of the fluorescence spot image of the sample is the basis and premise of digital PCR results processing.The specific fluorescence information of the sample can be separated from the digital PCR fluorescence image by using spectral unmixing technology,which can reduce the interference of non-sample fluorescence and background,improve the anti-interference ability of the whole system.In this paper,the spectral unmixing algorithm is applied to digital PCR microfluidic chip fluorescence image processing,development and use of spectrum unmixing algorithm to separate different fluorescence spectrum information in the digital PCR images.After the unmixing,the fluorescence information of the single spectrum emitted by the sample is obtained,and the fluorescence point image is reconstructed to exclude some interference of other fluorescence spectra and background,which is more conducive to the subsequent image processing and analysis and to improve the detection accuracy.The specific work and results of this paper are including:1.Various spectral unmixing algorithms are studied and analyzed,and the algorithm suitable for digital PCR fluorescence image is selected.The spectral unmixing algorithm(N-FINDR)for digital PCR fluorescence image is developed and implemented,and the results of spectral unmixing of two-color fluorescence image on microfluidic chip are obtained.Deal with the abundance and pixel error in the algorithm.2.The feasibility of the proposed method is verified by the experiment of two kinds of fluorescent dyes with different spectral characteristics.Then the experiment and the algorithm are replicated and verified on the microfluidic chip.After analyzing the chip images and data,it is proved that the microfluidic system is also suitable for this algorithm.3.The pre-processing technologies of fluorescence image are studied.RGB color model is used to calculate the fluorescence image,and the luminance value of the fluorescence point is obtained.And the luminance values under different spectral conditions are normalized.4.According to the characteristics of fluorescence image of digital PCR microfluidic chip,a localization algorithm of the brightest spot in a small area is proposed and used,and the characteristic values of fluorescence point location are extracted accurately.The theoretical basis of the new algorithm is analyzed and the accuracy of the results is verified.On the basis of point localizations,some interference is removed by the correction algorithm.The possible tilt angle of the image is corrected.The new application of spectral unmixing algorithm and the image preprocessing methods will improve the anti-background fluorescence interference ability of the total system and the accuracy of cut-off value setting in the future,so as to promote the development of digital PCR technology.