Karvelis P S, Fotiadis D I, Georgiou I, Sakaloglou P
Dept. of Computer Science, University of Ioannina, Ioannina, Greece, GR 45110.
Annu Int Conf IEEE Eng Med Biol Soc. 2009;2009:3601-4. doi: 10.1109/IEMBS.2009.5333757.
Color chromosome classification (karyotyping) allows simultaneous analysis of numerical and structural chromosome abnormalities. The success of the technique largely depends on the accuracy of pixel classification. In this paper we present a method for multichannel chromosome image classification based on support vector machines. First, the image is segmented using a multichannel watershed segmentation method. Classification of the pixels of the segmented regions using support vector machines is then employed. The method has been tested on images from normal cells, showing the improvement in classification accuracy by 10.16% when compared to a Bayesian classifier. The increased classification improves the reliability of the M-FISH imaging technique in identifying subtle and cryptic chromosomal abnormalities for cancer diagnosis and genetic disorders research.
彩色染色体分类（核型分析）可同时分析染色体的数目和结构异常。该技术的成功很大程度上取决于像素分类的准确性。在本文中，我们提出了一种基于支持向量机的多通道染色体图像分类方法。首先，使用多通道分水岭分割方法对图像进行分割。然后采用支持向量机对分割区域的像素进行分类。该方法已在正常细胞图像上进行了测试，与贝叶斯分类器相比，分类准确率提高了10.16%。分类准确率的提高提高了M-FISH成像技术在识别癌症诊断和遗传疾病研究中细微和隐匿性染色体异常方面的可靠性。
