Objective models of compressed breast shapes undergoing mammography

Abstract

Purpose: To develop models of compressed breasts undergoing mammography based on objective analysis, that are capable of accurately representing breast shapes in acquired clinical images and generating new, clinically realistic shapes. Methods: An automated edge detection algorithm was used to catalogue the breast shapes of clinically acquired cranio-caudal (CC) and medio-lateral oblique (MLO) view mammograms from a large database of digital mammography images. Principal component analysis (PCA) was performed on these shapes to reduce the information contained within the shapes to a small number of linearly independent variables. The breast shape models, one of each view, were developed from the identified principal components, and their ability to reproduce the shape of breasts from an independent set of mammograms not used in the PCA, was assessed both visually and quantitatively by calculating the average distance error (ADE). Results: The PCA breast shape models of the CC and MLO mammographic views based on six principal components, in which 99.2% and 98.0%, respectively, of the total variance of the dataset is contained, were found to be able to reproduce breast shapes with strong fidelity (CC view mean ADE = 0.90 mm, MLO view mean ADE = 1.43 mm) and to generate new clinically realistic shapes. The PCA models based on fewer principal components were also successful, but to a lesser degree, as the two-component model exhibited a mean ADE = 2.99 mm for the CC view, and a mean ADE = 4.63 mm for the MLO view. The four-component models exhibited a mean ADE = 1.47 mm for the CC view and a mean ADE = 2.14 mm for the MLO view. Paired t-tests of the ADE values of each image between models showed that these differences were statistically significant (max p-value = 0.0247). Visual examination of modeled breast shapes confirmed these results. Histograms of the PCA parameters associated with the six principal components were fitted with Gaussian distributions. The six-component model was also used to generate CC and MLO view mammogram breast shapes, using the mean PCA parameter values of these distributions and randomly generated values based on the fitted Gaussian distributions, which resemble clinically encountered breasts. A spreadsheet with the data necessary to apply this model is provided as the supplementary material. Conclusions: Our PCA models of breast shapes in both mammographic views successfully reproduce analyzed breast shapes and generate new clinically relevant shapes. This work can aid in research applications which incorporate breast shape modeling, such as x-ray scatter correction, dosimetry, and image registration.

X‐ray scatter correction in breast tomosynthesis with a precomputed scatter map library

“…This study utilizes the objective PCA model of patient breasts during acquisition of a cranio-caudal view mammographic image described by Feng et al 17 As the image of the compressed breast shape undergoing mammography is identical to the central angle (0 • ) projection of the compressed breast shape undergoing DBT, the mammography model forms the basis for the DBT model. The 3D breast volume (3DV) is built using the 2D breast shape (2DS) generated by the PCA model and an estimated curvature of the compressed breast volume.…”

“…To create this library, the parameters of the 4-component PCA model were selected according to the values in Table I. It has previously been shown that a 6-component PCA model results in greater accuracy of the modeled breast shape, 17 but the two additional latter components, which contribute the smallest amount of information to the shape, were deemed unnecessary for a low frequency signal such as x-ray scatter. This was confirmed by examination of the x-ray scatter maps resulting from five-and sixcomponent models.…”

“…The mean (μ) and standard deviation (σ ) of each principal component were used for these parameter selections. As the first component, α, has been shown to be the greatest determinant of the breast shape and size, 17 five values of α were chosen. For the other three components, only three values were chosen, since the differences introduced by these components on the shape of the breast, and therefore on the simulated scatter maps, were less pronounced.…”

“…To select the appropriate x-ray scatter map from the library, the central angle projection of each patient acquisition set was subjected to automated edge detection, as described by Feng et al 17 Automatic thresholding and breast tissue isolation, via the extraction of connected components, were performed to obtain a binary image of the breast tissue under examination. The breast shape edge was then located using Roberts cross gradient operators and edge thinning.…”

“…16 Here we present a study on the use of a library of precomputed DBT x-ray scatter maps with the scatter correction algorithm, derived from a previously developed model of clinically relevant compressed breast shapes based on principal component analysis (PCA). 17 The PCA-based breast shape model for the cranio-caudal (CC) view was used to generate compressed breast volumes that span the range of clinically encountered breast shapes and sizes. This study takes advantage of the fact that in DBT, the compressed breast is positioned in the same manner as in full field digital mammography (FFDM).…”