Malignancy prediction by using characteristic-based fuzzy sets: A preliminary study

Fuzzy sets approaches are widely used in medical diagnosis systems, since medical data is defined as uncertain, complex and subjective in most applications. In this paper we adapt and propose a fuzzy set based classification scheme for predicting malignancy of small pulmonary nodules. Our method basically consists of two steps: classification of nodule characteristics from image features and fuzzy set based inference engine for predicting malignancy. Results are compared with single classifiers which are trained on image features and with our previous methods. The results are promising for further development. Simple structure of the method makes it easier to implement and offers advantages over complex methods.