Breast tissue segmentation by fuzzy C-means

Abstract

Introduction Mammography is a worldwide image modality used in screening breast cancer. Due to its large availability, mammograms can be used to measure breast density. Women with high mammographic density have a four-to-sixfold increase in their risk of developing breast cancer. Therefore, studies have been made to accurately quantify mammographic breast density. In clinical routine, radiologist perform subjective image evaluations through BIRADS (Breast Imaging Reporting and Data System). Purpose The aim of this work was develop an automatic methodology to estimate the percentage of mammographic breast density using digital mammography. We used Fuzzy C-means Clustering (FCM) to segment fibroglandular and adipose tissues from breast mammography, using Matlab software. Materials and methods The algorithm uses FCM features (mean, standard deviation, kurtosis, entropy and others) to automatically segment tissues using mammograms. The mammographic breast tissue percentage was measured by the relation between fibroglandular tissue and the sum of fibroglandular and adipose tissues. The percentage was compared with the assessment made by radiologists using BIRADS system for each evaluated image. Results The comparison between methods shows 93% of concordance between the developed method and BIRADS system. The differences between methods, although small, were mainly attributed to subjective visual analysis made by radiologists. Conclusion The proposed method can automatically segment fibroglandular and adipose tissues with high performance. These results will be used in a complete work, which will estimate the volumetric breast density through digital mammography. The volumetric breast density will be used to calculate the mean glandular dose. Disclosure The authors declare that there is no conflict of interest.