Segmentation of Microscopic Breast Cancer Images for Cancer Detection

Breast cancer is one of serious diseases that affect mainly woman and late diagnosis can lead to death. However early diagnosis increases survivability significantly, therefore making it very important. There are different diagnosis techniques for early detection of breast cancer. Breast tissue samples analyzed under microscope is considered reliable way to diagnose breast cancer. Automated classification techniques are so popular in many areas in order to reduce human dependency considering third world countries. Our purpose is to determine if sample is malignant or benign in automated manner. Many algorithms are studied so far in medical area along with other areas. However, algorithms are generally too complex even for simple tasks. We propose a simple algorithm that can differentiate cancerous and non-cancerous samples from breast tissue in automated manner. The images were taken from Near East University Hospital which is consisted of 50 cancerous and 100 healthy images. Total of 150 images were correctly differentiated through our algorithm.