Automated classification of Pap smear tests using neural networks

The preliminary results of a project that automates the classification of Pap smear samples are given. In the preprocessing stage, first a set of ten features is extracted from a Pap smear image and is used to form the feature space. Then, the standard “The Bethesda System” (TBS) rules are translated into fuzzy rules that are used to classify the Pap smear test into normal and abnormal classes based on the extracted features. A feedforward neural network is applied for the sample for which fuzzy logic based classification is unclear. The high accuracy of classification of neural network on the preliminary results indicates the successful performance of the system