二维凝胶图像一致蛋白质点集提取方法

Method of extracting consistent protein spot set from 2-DE images

针对凝胶图像间蛋白质点人工比对工作量大和易出错的问题,提出一种基于粗配准—精匹配的一致蛋白质点集自动提取方法。根据凝胶图像的SIFT特征匹配结果对凝胶图像进行粗配准;采用标记分水岭方法进行点检测;再根据蛋白质点的距离相似度和灰度相似度进行双向匹配,提取一致蛋白质点集。通过对多种不同来源的凝胶图像进行实验,提取一致蛋白质点集的正确率高于97%。结果表明,该方法在具有缩放、旋转、亮度和噪声差异的凝胶图像中提取一致蛋白质点集都具有较好的稳定性。

In order to solve the problem that extracting consistent protein spot set in gel images by hand leads to heavy work- load and error-prone results, this paper proposed an intra-group consistent protein spot set extraction method based on image a- lignment and points match. Firstly,it coarsely aligned the intra-group gel images based on the matched SIFT feature points. Then,it detected the protein spots in the aligned images with label controlled watershed algorithm. At last,it match the protein spots bi-directionally based on distance similarity and gray similarity to extract the consistent protein spots set. Four experi- ments was carried on different 2-DE images and the correct rate was 97%. The results show that the proposed method is effec- tive and stable in 2-DE images with different characteristics such as zooming, rotation, different brightness, noise and so on.