Breast cancer is the second largest cause of cancer deaths among women. Mainly, this disease is tumor related cause of death in women. Early detection of breast cancer may protect women from death. Various computational methods have been utilized to enhance the diagnoses procedures. In this paper, we have presented the genetic algorithm (GA) based association rule mining method which can be applied to detect breast cancer efficiently. In this work, we have represented each solution as chromosome and applied to genetic algorithm based rule mining. Association rules which imply classification rules are encoded with binary strings to represent chromosomes. Finally, optimal solutions are found out by develop GA-based approach utilizing a feedback linkage between feature selection and association rule.
Top3.1 Introduction
Cancer data classification and clustering (Kharya, S. 2012) have been the focus of critical research in the area of medical and artificial intelligence. Health care is now days very important for human being. Breast cancer is the most common cancer in women in many countries. It is a malignant tumor that starts in the cells of the breast. A malignant tumor is a group of cancer cells that can grow into surrounding tissues or spread to distant areas of the body (Gandhi, K. R., Karnan, M., & Kannan, S. 2010, February). Various data mining techniques have been used to improve the diagnoses procedures and to aid the physician's efforts (Safavi, A. A., Parandeh, N. M., & Salehi, M. 2010). Screening mammography is the best tool available for detecting cancerous lesions before clinical symptoms appear (Gandhi, K. R., Karnan, M., & Kannan, S. 2010). Early detection and treatment of breast cancer can significantly advance the survival rate of patient. However, this is a challenging problem due to structure of cancer cells. Breast cancer detection, classification, scoring and grading of histopathological images is the standard clinical practice for the diagnosis and prognosis of breast cancer. It is a very complex and time-consuming duty for a pathologist to manually perform these tasks. Robust and efficient computer aided systems are therefore indispensible for automatic breast cancer detection.
In this chapter, we have considered classification problem as an association rule mining where antecedent part of a rule is considered as different conditions and consequent part of the rule is considered as class label. Data mining algorithms like association rule mining (Agrawal, R., & Srikant, R. 1994) perform an exhaustive search to find all rules satisfying some constraints. Hence, the number of discovered rules from database can be very large. Based on the earlier works, it is clear that it is difficult to identify the most effective rule. Therefore, in many applications, the size of the dataset is so large that learning might not work well before removing the unwanted features. To reduce the search space for large dataset, we have applied genetic algorithm over rule mining to make the learning efficient. Finally we have tried to find out best rules which can be applied for breast cancer detection (Anunciaçao, O., Gomes, B. C., Vinga, S., Gaspar, J., Oliveira, A. L., & Rueff, J. 2010).