A Hybrid Metaheuristic Approach for Automatic Clustering of Breast Cancer

Breast cancer is one of the most widely prevalent cancer in both men and women, however, it is far more common in women. Early diagnosis can help reduce the risk of metastasis to other organs and reduce death rates. Many research studies managed to cluster Breast cancer datasets and help in the accurate detection of tumors nevertheless a few ones could detect the number of clusters automatically. Clustering aims to partition the data into malignant or benign tumors. Recently researchers focused on automatic metaheuristic techniques reforming them to solve the clustering problem. In this paper, a new hybrid technique is proposed to determine the number of clusters with no need of prior information. The genetic algorithm is hybridized with cuckoo search algorithm to automatically cluster the Wisconsin Breast Cancer dataset. Also, a study among obtained results showed that the hybrid genetic cuckoo search algorithm outperformed the standard genetic algorithm and cuckoo search algorithm achieving an adjusted rand index of 0.84 and an adjusted mutual information of 0.74 and accuracy of 84 percent. Moreover, hybrid genetic cuckoo search algorithm was compared to the competing algorithms mentioned in the literature review showing a superior performance.