微阵列图像降噪和自动定位

Noise Reduction in Microarray Images and Automatic Addressing

针对基因芯片实验中存在的数据处理效率低、数据分析质量差等问题.将一种新的小波降噪技术和自动样点定位方法用于cDNA微阵列图像处理.研究表明:基于小波分解的降噪方法,对微阵列图像的噪声滤除非常有效,平均峰值信噪比可达到65dB.利用一种允许高吞吐量分析的全自动划格方法,对微阵列图像进行自动定位,对于基因表达微阵列图像精度可达到97%.

There are many factors which could cause error in microarray experiment. These could result in difference between ideal image and actual image. The efficiency of the manual-gridding method is low and the quality is not very good. In this paper, a microarray image enhancement of cDNA microarray images using a novel wavelets denoising technique and automatic addressing is presented. This study shows that robust denoising, based on wavelet decomposition and spot detection, can be used successfully to remove the noise from microarray images. The CPSNR is on the verge of 65dB. In addition, this paper introduces a fully automatic gridding method which permits a high throughput analysis, taking into account the subarray alignment problem. The method has the 97 percent accurate rate to effectively detect the grids on microarrays images of gene expression profiles.