The aim of this project is proposing a framework for analyzing the image information in mammograms as a function of the breast anatomy. Specifically, instead of using classical Cartesian coordinates, image intensity values of the segmented breast are re-mapped using standardized coordinates that depend on the breast anatomy. This new representation is then exploited in order to guide subsequent image analysis steps, namely feature extraction and classification. This framework takes into account the variability of the breast in terms of shape, size and orientation, and is independent to system-related variables, such as pixel size and spatial resolution. An important advantage of this approach is the feasibility for a straightforward integration with existing developments in each one of the blocks of the classical approach.

As illustrated below, the proposed anatomical breast mapping system works in three steps: 1) Contour detection, 2) Generation of the ST coordinate system, and 3) Anatomical breast mapping. The proposed mapping framework has applications in implicit breast image registration, automatic detection of region of interest (ROI), and anatomical breast image analysis. For further details, please see (Pertuz et al., 2014; Torres & Pertuz, 2016)

As illustrated below, the proposed anatomical breast mapping system works in three steps: 1) Contour detection, 2) Generation of the ST coordinate system, and 3) Anatomical breast mapping. The proposed mapping framework has applications in implicit breast image registration, automatic detection of region of interest (ROI), and anatomical breast image analysis. For further details, please see (Pertuz et al., 2014; Torres & Pertuz, 2016)

Figure: Anatomical breast mapping. From left to right: contour detection, generation of the ST coordinate system, and Anatomical breast mapping.