Automatic segmentation of breast and fibroglandular tissue in breast MRI using local adaptive thresholding

Abstract

Breast density is considered as an important risk factor associated with the development of breast cancer. Breast and fibroglandular tissue segmentation is the main step to compute breast density in Magnetic Resonance Imaging (MRI). This study presents an automatic algorithm to segment breast and fibroglandular tissue in MRI. It is a difficult task due to bias field and similar signal intensity between fibroglandular tissue and pectoral muscle. Our proposed segmentation approach has been developed based on the local adaptive thresholding to dominate on intensity inhomogeneity due to bias field and the low contrast intensity of the boundary between breast and pectoral muscle. The presented approach is validated with a dataset of 2520 bilateral axial breast MR images from 45 women that include all of Breast Imaging Reporting and Data System (BI-RADS) breast density range. Five quantitative metrics as Dice Similarity Coefficient (DSC), Jaccard Coefficient (JC), total overlap, False Negative (FN) and False Positive (FP) are employed to compare similarity between manual and automatic segmentations. For breast segmentation, the presented approach achieves DSC, JC, total overlap, FN and FP values of 0.90, 0.82, 0.89, 0.1 and 0.09, respectively. For fibroglandular tissue segmentation, we attain DSC, JC, total overlap, FN and FP values of 0.96, 0.94, 0.98, 0.02 and 0.04, respectively.