Morphometric Pattern Analysis of Basal Cell Nuclei for Oral Cancer Screening

This work presents a quantitative approach for discrimination of Oral Submucous Fibrosis (OSF) to Normal Oral Mucosa (NOM) in respect to size and shape properties of the basal layer, first layer in epithelium. Practically, basal cells form the proliferative compartment of the epithelium, and therefore changes in the morphometry of basal cells may have serious implications on future cell behavior, including malignant transformation according to onco-pathologists vision. In view of this, the changes in shape and size of the nuclei in the basal cell layer of the oral epithelium have been studied here by developing an automated image analyzer. Geometric, Zernike moments and transformation based features are extracted for morphometric pattern analysis of the nuclei. These features are statistically analyzed along with 3D visualization in order to discriminate the groups. Results showed increase in the dimensions (area and perimeter) and shape parameters of the nuclei from normal mucosa to OSF with dysplasia. Finally, pattern analyzer is employed using Bayesian approach and error back-propagation neural network. The performance is evaluated by partitioning the whole data set into various combinations of training-testing subsets, finally which converge to overall accuracies 97.02% for neural network and 97.93% for Bayesian classifier respectively.