Validation of Mammographic Texture Analysis for Assessment of Breast Cancer Risk

摘要

The purpose of this study was to evaluate, using full-field digital mammography (FFDM), our prior computerized texture analysis method that was developed on screen-film mammography method. The evaluation included the analyses on the parenchymal patterns of women with BRCA1 or BRCA2 gene mutations and of women at low risk of developing breast cancer. A total of 180 cases, including 80 women with BRCA1 or BRCA2 gene mutations and 100 low-risk women, were retrospectively collected under an institutional review board approved protocol. Images were obtained with a GE Senographe 2000D FFDM system with 0.1 mm pixel size and 12-bit quantization. Regions-of-interest (ROIs), 256 pixels by 256 pixels in size, were manually selected from the central breast region immediately behind the nipple. The ROIs were used in subsequent computerized feature extraction to assess the mammographic parenchymal patterns in the images. Various mammographic parenchyma features based on local composition, gray-level histogram analysis, spatial relationship among gray-levels, fractal analysis, edge frequency analysis, and Fourier analysis, were automatically extracted from these ROIs. Receiver Operating Characteristic (ROC) analysis was used to assess the performance of the computerized texture features in the task of distinguishing between gene-mutation carriers and low-risk subjects. Computerized texture analysis on digital mammograms demonstrated that gene-mutation carriers and low-risk women have different mammographic parenchymal patterns. In addition, in a round-robin-by-case evaluation with the FFDM dataset with linear discriminant analysis, an AUC value of 0.88 was obtained. Our results indicate the transferability of these radiographic biomarkers for breast cancer risk assessment from SFM to FFDM.