Breast Cancer Diagnosis using Digitized Images of FNA Breast Biopsy and Optimized Neurofuzzy System

Breast cancer is the second highest cause of cancer death among women all around the world, especially between the ages of 40 and 65. Early detection of breast cancer tumor type, benign or malignant, has remained a big challenge for physicians and radiologists. By timely identifying the tumor and knowing its type, it can allow physicians to conduct the treatment in the right direction and to take appropriate medications for that type of tumor. Therefore, it is highly important to know the type of tumor accurately to save the patient's life. This paper proposes an intelligent diagnosis method for breast cancer tumor type detection and classification based on fine needle aspirate (FNA) of a breast mass and optimized fuzzy sets. The developed method applied image processing and feature selection techniques to analyze the digitized image of a FNA. In the developed method, nine features computed from a digitized image of a FNA of a breast mass are uses as the initial input of neuro-fuzzy system. In the proposed method, we applied the Association Rules (AR) to find the most informative and effective features of the extracted features. Moreover, in order to advance the act of neuro-fuzzy system, Chaotic Bat Optimization Algorithm (CBOA) is used as learning algorithm. The proposed method is tested on Breast Cancer Wisconsin (Diagnostic) data set. The obtained numerical analysis showed that the developed diagnostic system has excellent performance and high degree of precision.