摘要

The identification of breast cancer genes can benefit highly from studies pertaining to the genetic effects on normal growth and morphogenesis of the mammary gland. Such studies currently lack, but need, standardized, quantitative assessment of relevant developmental features. To address this need, we created two computational frameworks for automated analysis of images of whole-mounted, carmine-stained murine mammary glands. The first framework is designed to quantitatively assess fat pad dimensions and size, percentage of epithelial filling of the fat pad, epithelial density within the fat pad area occupied by the epithelium, and longitudinal and lateral extension of the epithelium into the fat pad; from images of whole glands. The second framework uses images of higher magnification and resolution of the ductal system to determine the number of end- and branching points, and ductal length and width. Our frameworks return the quantitative data together with quality control (QC) images to the user, for verification of correct segmentation of the epithelium and fat pad, and of correct identification of epithelial ducts and their branching and end points: We quantitatively tested the sensitivity of our frameworks to differences in exposure conditions during image acquisition and to intra- and inter-user variation of image analysis. These analyses revealed that our frameworks are accurate and robust. Combined, these frameworks form an excellent tool for studies of mammary gland development. We will make this tool available as software called MammoQuant, to facilitate quantitative and standardized assessment of mammary gland development in the context of different gene mutations.

  • 出版日期2012-12

全文