Geometry-Based Counting and Classification of WBCs for Analysis of Leukocyte Disorders

White blood cells (WBCs) or leukocytes represent an important component of the immune system that serve as a defence mechanism against infectious diseases. Total leukocyte counts and ratios of its sub-types, e.g., neutrophil–lymphocyte ratio, neutrophil–monocyte ratio, etc. are important indicators for diagnosis of various diseases. The problem of an accurate count of WBCs can be automated with the help of proper cell segmentation, feature extraction, and classification. Several complex models have been proposed for the same purpose. This paper discusses an efficient classification technique for the classification and counting of WBCs. In this approach, the geometry-based features are extracted as part of prepossessing. A new feature (number of concave points) is also introduced with the help of an improved concavity detection algorithm. Finally, extracted features are used for classification to the support vector machine (SVM) classifier. The proposed algorithm is executed on leukocyte images from Mendeley and LISC datasets. The test results provide a good accuracy level.