Tabrizi P R, Rezatofighi S H, Yazdanpanah M J
Control and Intelligent Processing Center of Excellence (CIPCE), school of Electrical and Computer Engineering, University of Tehran, P.O. Box 14395/515, Iran.
Annu Int Conf IEEE Eng Med Biol Soc. 2010;2010:5593-6. doi: 10.1109/IEMBS.2010.5626788.
Designing an effective classifier has been a challenging task in the previous methods proposed in the literature. In this paper, we apply a combination of feature selection algorithm and neural network classifier in order to recognize five types of white blood cells in the peripheral blood. For this purpose, first nucleus and cytoplasm are segmented using Gram-Schmidt method and snake algorithm, respectively; second, three kinds of features are extracted from the segmented areas. Then the best features are selected using Principal Component Analysis (PCA). Finally, five types of white blood cells are classified using Learning Vector Quantization (LVQ) neural network. The performance analysis of the proposed algorithm is validated by an expert's classification results. The efficiency of the proposed algorithm is highlighted by comparing our results with those reported in a recent article which proposed a method based on the combination of Sequential Forward Selection (SFS) as the feature selection algorithm and Support Vector Machines (SVM) as the classifier.
在文献中先前提出的方法中，设计一个有效的分类器一直是一项具有挑战性的任务。在本文中，我们应用特征选择算法和神经网络分类器的组合，以便识别外周血中的五种白细胞类型。为此，首先分别使用Gram-Schmidt方法和蛇形算法分割细胞核和细胞质；其次，从分割区域中提取三种特征。然后使用主成分分析（PCA）选择最佳特征。最后，使用学习向量量化（LVQ）神经网络对五种白细胞类型进行分类。所提算法的性能分析通过专家的分类结果进行验证。通过将我们的结果与最近一篇文章中报道的结果进行比较，突出了所提算法的效率，该文章提出了一种基于顺序前向选择（SFS）作为特征选择算法和支持向量机（SVM）作为分类器的组合方法。
