Applications Of Digital Image Processing To Surface Plasmon Resonance Microscopy Imaging

As the chemical science gradually extends to the microscopic scale,a variety of microscopy imaging techniques with excellent temporal and spatial resolution have been widely and deeply applied to research providing a lot of space and time information about the surface morphology and chemical composition of the samples.At the same time,the information science and space science have already established many unique high-dimensional data analysis methods,which can effectively extract hidden information from a large number of images and video data.In this thesis,the image data obtained by the surface plasmon resonance microscopy(SPRM)are taken as the research object,and various digital image processing techniques are introduced to discover the information hidden behind these data.The main contents are as follows:1.Determination of Absolute Concentration of Gold Nanoparticles Based on Image Recognition Algorithm.The molar concentration of gold nanoparticles must be measured accurately in order to apply gold nanosols widely in the fields of sensing,diagnosis and nano-pharmaceutics.At present,most of the methods for determining the concentration of gold nanoparticles are only applicable to particles with regular shape and narrow particle size distribution.In this thesis,a method for determining the absolute concentration of gold nanoparticles based on image recognition algorithm is proposed.The algorithm firstly filters the black and white noise in the pictures using the median filter,then performs two-dimensional convolutionoperation on the pictures to make the signal peak conform to the Gaussian distribution,and finally locates the local brightest point in the pictures to confirm the position and number of nanoparticles.This study established a computer program based on this algorithm.By counting the number of thousand gold nanoparticles colliding onto the sensor chip in a certain period of time,we get the curve of the collision frequency of the nanoparticles decreased with time.This curve is then combined with the Stokes-Einstein equation to calculate the absolute concentration of gold nanoparticles without the need to establish a working curve2.Visualizing the Process of Collision to Determine the Absolute Concentration of Polystyrene Nanoparticles.Polystyrene nanoparticles can be used as carriers for a variety of drugs.However,polystyrene nanoparticles of high concentration is toxic,so it is necessary to accurately determine their molar concentration during drug development.In this thesis,a method to directly determine the absolute concentration of polystyrene nanoparticles without any prior knowledge and working curve was proposed.This method combined the curve of collision frequency decreasing vs time obtained by image recognition algorithm with the diffusion coefficient obtained by the visualization of collision process to determine the absolute concentration of polystyrene nanoparticles.In this method,the additional operation of measuring the diffusion coefficient by a dynamic light scattering instrument is avoided by fitting the collision process with a function to directly calculate the diffusion coefficient of the polystyrene nanoparticles.Compared with the previous chapter's research,the requirements for prior knowledge of the samples are further reduced.3.Cell Viability Assay Based on Image Correlation Spectroscopy(ICS).Cell viability assay is commonly used in drug screening to save human and material resources during the drug development period.However,the commonly used cell viability assay is based on the overall response of a large number of cells,unable to observe the dynamic changes of individual cells in real time.SPRM is able to detect the binding kinetics of drug-receptor in single cells but unable to reflect the viability of cells in real time,owing to the lack of method to analyze the SPRM signals produced by cell motility.These signals are generated at random positions and changing rapidly.In this thesis,a data analysis algorithm is proposed,which can extract characteristic signals from these random SPRM signals to realize the viability analysis of single cell.In addition,this method overcomes the effect of the system drift on the experimental results.The method was applied to the study of HeLa cell viability and successfully measured the cell viability decay kinetics during the process of single living cell fixation.4.Drift Correction Algorithm Based on Pixel Reconstruction.The spatial resolution of the microscopy is increasing and the sampling rate is getting faster.Observation scale is improved from micron-level to nano-level.However,the impact of system drift has been magnified at the same time.When we use a microscope to capture the images of a sample for a long time,the images may drift due to the fluctuation of temperature or mechanical slack.In this thesis,a general adaptive algorithm based on pixel reconstruction is proposed for the correction of image horizontal drift.The drift correction algorithm performs cross-correlation calculation on the entire picture and extracts the average drift information of the entire images.Therefore,the errors caused by the randomly manual selection of marker points and personal preference are excluded.The drift information extracted from the entire images also avoids the errors caused by the random movement of the selected single marker.This algorithm does not need to transform the instrument or carry on special operation in the sampling process.Experimental data with horizontal drift can be corrected using this method.This study shows the effectiveness and reliability of the drift correction algorithm using fluorescence microscopy images and SPRI image data as examples.