Fuzzy Mathematical and Shape Theoretic Approach to Cervical Cell Classification
Lipi B. Mahanta, Dipen Nath, Debaleena Majumdar, Sainesh Karan, Jitendra Sharma · International Journal of Computer Applications · 2011
ABSTRACT We applied traditional fuzzy mathematical approach with enhanced initialization procedure to segment Pap smear images of cervical cells. The segmented images of the cervical cells were analyzed with the help of shape theory to classify them accordingly to the presence of abnormality in the morphological behavior of the cells. General Terms Fuzzy mathematical approach, Shape theory. Keywords Cervical cells, Pap smear, Fuzzy c-means, nucleus and cytoplasm. 1. INTRODUCTION Automated classification of cervical cells is a challenging task in the field of medical sciences. Classification of cervical cells is the preliminary criteria for finding any abnormality occurring in the cervical region. Manual observation of cells in Pap smear slide is prone to error due various factors such huge number of cells in a slide, fatigue and tidiness in observation of large number of slides, different expertise level of the health personnel etc. An automated intelligent method tends to give uniform and accurate result. Feeding of morphological and other information regarding the cells into various intelligent methods is the common paradigm of several researchers in automated cell classification process. Morphological information of the cells like size and shape are computed in order to create the dataset for classification purpose. A set of information regarding the morphology of the cells and colour intensity is computed from the Pap smear images using commercially available image analysis software and is classified using traditional classifiers such Minimum Distance classifier and advanced classifier such as Nearest Neighbour with GA Feature Selection, Nearest Neighbour with TABU Search Feature Selection and Ant Colony optimization [11].