A typical lymphocyte cells identification based combination of similarity measurement and affine geometric transformation

Although morphological examination of peripheral blood smear provides significant diagnostic information of blood diseases, considerable development is required to facilitate the examination procedure. In this study, the identification of atypical lymphocyte cells is considered for the diagnosis of infectious and non-infectious diseases, including immunological and viral infections. In clinical laboratories, a digital camera is utilized for microscopy imaging scenarios, by taking captions from blood film. Hematologist’s human-based diagnoses are considered in designing the proposed algorithm. For blood cell counting, the color contrast between white blood cells (WBCs) and Red Blood Cells (RBCs) is employed in terms of the grayscale threshold. In addition, the type of WBCs is identified by combining both 2D cross correlation and affine geometric transformation. Results show the ability of the proposed algorithm to differentiate the atypical lymphocyte cells at good enough color contrast, where the correlation coefficient increased from 0.54 to 0.79 according to affine transformation of rotation.