A novel clustering method based on K-MEANS with region growing for micro-calcifications in mammographic images

Abstract

Breast cancer is one of the most dangerous malignant tumors of women in the world. A particularly important clue of such disease is the presence of clusters of micro-calcifications. However, it is difficult for radiologists to provide both accurate and uniform evaluation for benign or malignant pathologic modifications of micro-calcifications. The radiologists are usually obtained by using human expertise in recognizing the presence of given patterns and types of microcalcifications. In order to automatically detect such clusters and improve the accuracy, in this paper, K-MEANS-based region growing clustering algorithm is proposed to automatically finding clusters of micro-calcifications in the phase of clustering in mammography. The approach has been successfully tested on a standard database of 30 mammographic images, publicly available.

Convolutional neural network for breast cancer diagnosis using diffuse optical tomography

“…Thus far, several studies have applied computer-aided diagnosis (CAD) techniques to breast cancer detection; these techniques include the use of artificial neural networks [13][14][15], fuzzy logic [16,17], Bayesian networks [18,19], decision trees [20,21], and k-means clustering [22,23]. However, few researchers have implemented CAD methods with DOT to diagnose breast cancer.…”