Abstract
Accurate segmentation of tissues is a prerequisite for a number of quantitative investigations of breast such as biomechanical modeling. A characteristic of breast that might affect the segmentation accuracy is the distribution of tissues. An appropriate method to classify breast with sparse distribution of fibroglandular tissue has so far not been studied.
Our aim is to identify a texture feature that can be used to classify the distribution of tissue. This work studied and evaluated texture features to classify fibroglandular tissue distribution in the breast. Radiotherapy planning computed tomography (CT) data of 24 patients in the supine position were analyzed. These breast datasets were visually assessed and ranked by an expert for the sparseness of the fibroglandular tissue distribution. In this study texture features were derived using properties of the radial glandular fraction (RGF) such as the mean, and standard deviation. RGF, presented by Huang et al., describes the distribution of fibroglandular tissue within the breast. Patients were grouped into non-sparse and sparse tissue distribution based on expert rank and texture features of these two groups were compared. A Wilcoxon rank test was used to test significance of differences between the two groups and the Bonferroni correction was used to adjust the significance threshold for multiple comparisons.
The radial glandular fraction of the two groups, non-sparse and sparse was substantially different for the three breast regions studied. There were significant differences between these two groups for a number of features. This study indicates that texture features derived from the RGF can provide a good indicator for the classification of the distribution of fibroglandular tissue in the breast. Performance of these texture features to classify the fibroglandular tissue distribution is presently being investigated.