Incorporating Robustness to Imaging Physics into Radiomic Feature Selection for Breast Cancer Risk Estimation

SIMPLE SUMMARY: Mammographic density estimates can be combined with radiomic texture features to offer an even better assessment of breast cancer risk. However, some feature variations will be due to true parenchymal differences between women, but others will be due to imaging physics effects (contrast, noise, and image sharpness); features robust to imaging physics effects should better model risk. To investigate this, we imaged an anthropomorphic phantom at various x-ray technique settings, allowing us to directly measure the effects of imaging physics on feature values. We compared these variations, for each feature, with the inter-woman variation in a screening population (552 cancer-free women) and the intra-woman variation between each woman’s left and right breasts, to assess which features were relatively robust to physics settings. We then tested more- versus less-robust features in modeling cancer risk on an independent case-control data set, and demonstrated that more-robust features were indeed better at risk prediction. ABSTRACT: Digital mammography has seen an explosion in the number of radiomic features used for risk-assessment modeling. However, having more features is not necessarily beneficial, as some features may be overly sensitive to imaging physics (contrast, noise, and image sharpness). To measure the effects of imaging physics, we analyzed the feature variation across imaging acquisition settings (kV, mAs) using an anthropomorphic phantom. We also analyzed the intra-woman variation (IWV), a measure of how much a feature varies between breasts with similar parenchymal patterns—a woman’s left and right breasts. From 341 features, we identified “robust” features that minimized the effects of imaging physics and IWV. We also investigated whether robust features offered better case-control classification in an independent data set of 575 images, all with an overall BI-RADS(®) assessment of 1 (negative) or 2 (benign); 115 images (cases) were of women who developed cancer at least one year after that screening image, matched to 460 controls. We modeled cancer occurrence via logistic regression, using cross-validated area under the receiver-operating-characteristic curve (AUC) to measure model performance. Models using features from the most-robust quartile of features yielded an AUC = 0.59, versus 0.54 for the least-robust, with p < 0.005 for the difference among the quartiles.