WBCs Segmentation and Identification System using PCA and PNN
Numerous diseases can be diagnosed based on the number of cells for each class of White Blood cells (WBCs) in the blood, therefore extracting information about that is considered very important for hematologists. the types of WBCs are Neutrophil, Eosinophil, Basophil, Monocyte and Lymphocyte; Each one of these classes is different from the other in the size and the shape of nucleus. This paper aimed to present automatic medical diagnostic system has the ability to segment the WBC from other components in microscopic blood images (95.65% accuracy rate of segmentation process) and then identify each type of WBCs using Principle Component Analyses (PCA) which is used for feature extraction operation then these extracted features are applied to the Probabilistic Neural Network (PNN) to classify the extracted features to the five classes of WBCs (94% accuracy rate of identification process). This system is implemented under the MATLAB® (version R2012b) platform.