Research And Implementation Of Karyotype Analysis System For Image Of Human Chromosome

Chromosome is a specific form formed by the genetic material of eukaryote during division and the carrier of genetic information,which is mainly composed of DNA and protein.It has the function of storing and transmitting genetic information.Karyotype refers to an image formed by arranging all chromosomes in a cell according to their size and shape characteristics.Karyotype analysis is a classical method of genetic analysis technology,and is one of the important means of biological genetic science research and medical genetic disease diagnosis.The early human chromosome karyotype analysis needs to be done manually by people with professional skills,which requires high quality of operators,time-consuming,laborious and inefficient,and can not meet the needs of scientific research and disease diagnosis.Recently,with the update of computer technology,the development of digital image processing theory and pattern recognition theory,computer-aided karyotype analysis has been realized,greatly improving the efficiency of karyotype analysis,and bringing convenience to the development of biology and medicine.Based on G-banding chromosome images,this paper uses digital image processing theory and pattern recognition technology to design algorithm function modules as well as combining with computer program design method to develop a complete set of chromosome image karyotype analysis software system to realize computer-aided karyotype analysis which aims to reduce the cost of chromosome karyotype analysis and improve the efficiency of chromosome karyotype analysis.Based on theoretical analysis and experimental data,this paper designs an image preprocessing method with Otsu image segmentation,morphological smoothing and morphological boundary extraction as the main processes in the chromosome image preprocessing stage.The length,centromere position and striation distribution of chromosomes are taken as the main characteristic data of chromosome images.In the process of extracting the characteristic data,for the extraction of the axis of chromosome images,a method is designed to extract the chromosome skeleton points by calculating the gray maximum of different gradient directions in the chromosomal distance transform images.In the process of centromere localization,the length of perpendicular line and contour transversal of each point of chromosome axis is proposed to represent the width of chromosome,and the minimum value of width is calculated to determine the position of centromere.In the banding banding distribution,the gray mean value in the direction of vertical axis is counted to quantify the banding distribution.According to the karyotype characteristics of human chromosomes and the basis of chromosome grouping,the extracted chromosome feature data were used to design a secondary classification model of intra-group classification based on the preliminary grouping of BP neural network and k-means clustering method.Finally,with the help of OpenCV library,Qt interface framework and other software tools,the method flow of this paper is integrated to design and implement the human chromosome image classification software system.The experimental data show that the system has achieved good automatic classification rate and automatic classification accuracy.