Research On Gene Chip Image Segmentation Algorithm Based On Deep Learning

Gene chip is a microarray composed by thousands of gene points,it is a technology to analyze gene expression.Gene expression can not only explore the potential genetic causes of human diseases,but also can be used in environmental hygiene research,drug discovery&development,clinical diagnosis&treatment,and cancer detection.Gene chip image processing is an extremely important part in gene expression analysis,the result of processing directly affect the accuracy and precision.Image processing mainly includes preprocessing,grid location,target segmentation,signal extraction and analysis.The main work of this paper is focused on the following points:First of all,the characteristics of different gene points were analyzed by gene chip image preprocessing,then gray transformation,contrast enhancement,filter and denoise processing are carried out,based on that do grid location.The grid location can be used to separate the target signal area,which lays the foundation for the target segmentation of the gene chip.Secondly,this paper compares the image segmentation algorithm based on clustering&classification,which include K-means,fuzzy local information C-means and support vector machine.Through the analysis of above-mentioned algorithms'advantages and disadvantages,introduces improved K-means algorithm,detailed introduces the fuzzy local information C-means algorithm,explains the application of support vector machine in image segmentation of gene chip images,support vector machine is how to obtain the gene chip images' features,and then leads to deep learning method.Finally,based on the summary of the above algorithm,the segmentation algorithm based on deep learning is proposed:constructing a framework of convolutional neural network,creating the simulated and real datasets,and through the two data sets of testing,a series of experiments on the effectiveness and robustness of the algorithm have been done.The experimental results show that the performance of the proposed algorithm is much better than that of K-means and fuzzy local information C means,Segmentation accuracy is higher than support vector machine...