Skip to Main Content
A system based on a fuzzy finite state machine (FFSM) has been developed for evaluating cytological features derived directly from a digital scan of breast fine needle aspirate (FNA) slides. The system uses computer vision techniques to analyse cell nuclei in order to extract determinate features and to try to find, by means of genetic algorithms (GA), the ideal FFSM that is able to classify them. This application to breast cancer diagnosis uses the characteristics of individual cells to discriminate benign from malignant breast lumps. In our system, we try to find a texture measurement that can be included in the feature set in order to improve the classifier performance: a complexity measurement of the structural pattern is used to discriminate between benign and malign cells. With this measure and the technique described, we have observed that not only is the absolute complexity of the image relevant, but also the way in which the complexity is distributed at different scales.