1st International and 10th National Iranian Conference on Bioinformatics

Diagnosis of breast cancer using K-Means and fuzzy C-Means clustering

Paper ID : 1118-ICB10

Authors:

Hamidreza Erfanian *¹, زهرا غفاری²

¹گروه بیوانفورماتیک، دانشکده علوم و فناوریهای نوین زیستی، دانشگاه علم و فرهنگ، تهران، ایران

²گروه بیوانفورماتیک، دانشگاه علم و فرهنگ، تهران، ایران

Abstract:

In recent years, breast cancer has been one of the most common causes of death among women [1]. Thermography is one of the fastest, cheapest, risk-free, radiation-free, and painless diagnostic methods available for this cancer [2, 3]. The use of new methods in image processing and machine learning has led to the use of thermographic images to successfully conduct studies to establish breast cancer diagnostic systems [4-8]. In this study, the diagnosis of breast cancer has been studied using K-Mean and fuzzy C-Mean clustering and an intelligent method has been used to separate healthy from unhealthy tissue and to separate mass in unhealthy tissue. In this method, clustering was performed using two methods: K-Mean and fuzzy C-Mean, then the cluster with the highest center intensity as the input of the area growth algorithm and the brightest pixel as the grain point of the area growth method were selected. The suspected area was determined according to the growth algorithm of the area and then the specificity of the suspected area was extracted based on the coefficient matrix. At this stage, the threshold was set for the four properties of the co-occurrence matrix and based on it, a decision was made about the suspicious area. Using this method, an intelligent system was designed that reduces the amount of human error in the diagnosis of cancer and will be able to detect the mass in the early stages of breast cancer with 91.67% classification and 89.65% sensitivity.

Keywords:

Breast cancer; Image processing; K-Means clustering Method; Fuzzy C- Means clustering Method.

Status : Paper Accepted (Poster Presentation)