Breast Cancer Risk Assessment by a Hybrid Interval Type-2 Fuzzy Cognitive Map Method

Abstract

This paper proposes a new method for accessing the breast cancer risk called Hybrid High-order Interval Type-2 Fuzzy Cognitive Map (H-HIT2 FCM). In a simple Fuzzy Cognitive Map (FCM), the weights between nodes and activation functions are constant in each iteration. As an extension in the high order FCM, each node has a different transformation function to make it more flexible. However, using FCM or high order FCM can not make a favorable response in uncertain situations. Applying type-2 Fuzzy Cognitive Map to obtain the weights of FCM, the resulted method will have much better responses in such uncertain situations. An H-HIT2 FCM is proposed in this work, assessing breast cancer risk in three modes of optimistic, realistic, and pessimistic. The proposed method has three levels. In the first level, the patient's profile, family history, and the inherited factors are tested by high order FCM. In the second level, by examining the mass characteristics obtained from the mammograms, the disease risk is achieved by high-order interval type-2 FCM in three modes of optimistic, realistic, and pessimistic. The exact position of the tumor is obtained in the third level. Finally, a Support Vector Machine predicts an overall breast cancer risk. Moreover, compared to the existing methods, the accuracy of the results is desirable. The three-mode assessment will help the patients and their physician run the best treatment. The proposed method is successfully tested on a real radiology dataset, and the corresponding results are reported.