Adaptive Intelligent Systems for Recognition of Cancerous Cervical Cells Based on 2D Cervical Cytological Digital Images

Abstract

To date, cancer of uterine cervix is still a leading cause of cancer-related deaths in women in the world. Papanicolau smear test is a well-known screening method of detecting abnormalities of the cervix cells. Due to scarce number of skilled and experienced cytologists, the screening procedure becomes time consuming and highly prone to human errors that leads to inaccurate and inconsistent diagnosis. This condition increases the risk of patients who get HPV infection not be detected and become HPV carriers. Coping with this problem, an adaptive intelligent system is developed to enable automatic recognition of cancerous cells from. Here pattern recognition is done based on three morphological cell characteristics, i.e. size, shape, and color features, and measured as numerical values in terms of N/C ratio, nucleus perimeter, nucleus radius, cell deformity, texture heterogeneity, wavelet approximation coefficients, and gray-level intensity. Through a supervised learning of multilayer perceptron network, the system is able to percept abnormality in the cervix cells, and to assign them into a predicted group membership, i.e. normal or cancerous cells. Based on thorough observation upon the selected features and attributes, it can be recognized that the cancerous cells follow certain patterns and highly distinguishable from the normal cells.