A Comprehensive Methodology for Determining the Most Informative Mammographic Features

Abstract

This study aims to determine the most informative mammographic features for breast cancer diagnosis using mutual information (MI) analysis. Our Health Insurance Portability and Accountability Act-approved database consists of 44,397 consecutive structured mammography reports for 20,375 patients collected from 2005 to 2008. The reports include demographic risk factors (age, family and personal history of breast cancer, and use of hormone therapy) and mammographic features from the Breast Imaging Reporting and Data System lexicon. We calculated MI using Shannon's entropy measure for each feature with respect to the outcome (benign/malignant using a cancer registry match as reference standard). In order to evaluate the validity of the MI rankings of features, we trained and tested naïve Bayes classifiers on the feature with tenfold cross-validation, and measured the predictive ability using area under the ROC curve (AUC). We used a bootstrapping approach to assess the distributional properties of our estimates, and the DeLong method to compare AUC. Based on MI, we found that mass margins and mass shape were the most informative features for breast cancer diagnosis. Calcification morphology, mass density, and calcification distribution provided predictive information for distinguishing benign and malignant breast findings. Breast composition, associated findings, and special cases provided little information in this task. We also found that the rankings of mammographic features with MI and AUC were generally consistent. MI analysis provides a framework to determine the value of different mammographic features in the pursuit of optimal (i.e., accurate and efficient) breast cancer diagnosis.

Developing a clinical utility framework to evaluate prediction models in radiogenomics

“…For each SNP, we quantified how many risky alleles were present (0, 1, or 2 risky alleles) as the value. For mammography findings described in Breast Imaging-Reporting and Data System (BI-RADS) lexicon 29 , we collected 5 most informative mammographic features 14 (Table 1). …”

“…However, early attempts to use genetic variants to predict breast cancer risk have demonstrated only modest improvements over conventional demographic risk factors 2, 8, 10 . On the other hand, there is a long history of risk estimation for breast cancer by using imaging findings garnered from mammograms 11–14 . However, it has proven a difficult task to characterize the molecular properties of malignancies using imaging findings.…”

Quantifying predictive capability of electronic health records for the most harmful breast cancer

“…A series of prediction models have been developed to predict the probability of malignant versus benign tumors, including demographic risk factors 4 , genetic variants 5–7 and imaging features 8–11 . However, few manuscripts document models to predict the “most harmful” breast cancers.…”

Developing a utility decision framework to evaluate predictive models in breast cancer risk estimation

“…34 After extraction, every mammographic feature takes the value "present" or "not present." From these features, we selected the most predictive abnormality descriptors based on the literature: 13 mass margin, microcalcification shape, microcalcification morphology, and architectural distortion. For microcalcification features, we consolidated the suspicious morphology descriptors (linear, amorphous, and pleomorphic) and suspicious distribution descriptors (clustered, segmental, linear) into the "present" category; cases lacking any of these descriptors in their records were assigned to the "not present" category.…”

“…On the other hand, there is a long history of risk estimation for breast cancer by using imaging findings. [10][11][12][13] Now, it is widely agreed that imaging findings, in concert with genetic variants will likely be necessary for accurate assessment of a patient's breast cancer risk. A promising new paradigm, "radiogenomics," delves into the analysis of the interaction of imaging findings and genetic variants for estimating cancer risk.…”