Classification of mammographic breast density and its correlation with BI-RADS in elder women using machine learning approach.

INTRODUCTION: Mammographic breast density (MBD) is a known risk factor for breast cancer and older women have higher incidence rates of breast cancer occurrence. The Breast Imaging Reporting and Data System (BI-RADS) is a commonly used MBD classification tool for mammogram reporting. However, they have limitations since there are reading inconsistencies between different radiologists with the visual assessment of breast density. METHODS: Digitised film-screen mammographic images were extracted from the Digital Database for Screening Mammography (DDSM). A machine learning project was developed using commercially available software with several predictive models applied to classify different amount of MBD on mammograms into different density groups. The effectiveness of different predictive models used in classifying the mammograms were tested by receiver operator characteristics (ROC) curve with comparison to the gold standard of BI-RADS classification. RESULTS: Three predictive models, Decision Tree (Tree), Support Vector Model (SVM) and k-Nearest Neighbour (kNN) showed high AUC values of 0.801, 0.805 and 0.810 respectively. High AUC values for the three predictive models indicates that the accuracy of the model is approaching that of the BI-RADS method. DISCUSSION: Our machine learning project showed to have capabilities to be potentially used in the clinical settings to help categorise mammograms into extremely dense breasts (BI-RADS Group A) from entirely fatty breasts (BI-RADS Group D). CONCLUSION: Findings from the present study suggest that the machine learning method is potentially useful to quantify the amount of MBD in mammograms.

Spatial Distribution and Quantification of Mammographic Breast Density, and Its Correlation with BI-RADS Using an Image Segmentation Method.

(1) Background: Mammographic breast density (MBD) and older age are classical breast cancer risk factors. Normally, MBDs are not evenly distributed in the breast, with different women having different spatial distribution and clustering patterns. The presence of MBDs makes tumors and other lesions challenging to be identified in mammograms. The objectives of this study were: (i) to quantify the amount of MBDs-in the whole (overall), different sub-regions, and different zones of the breast using an image segmentation method; (ii) to investigate the spatial distribution patterns of MBD in different sub-regions of the breast. (2) Methods: The image segmentation method was used to quantify the overall amount of MBDs in the whole breast (overall percentage density (PD)), in 48 sub-regions (regional PDs), and three different zones (zonal PDs) of the whole breast, and the results of the amount of MBDs in 48 sub-regional PDs were further analyzed to determine its spatial distribution pattern in the breast using Moran's I values (spatial autocorrelation). (3) Results: The overall PD showed a negative correlation with age (p = 0.008); the younger women tended to have denser breasts (higher overall PD in breasts). We also found a higher proportion (p < 0.001) of positive autocorrelation pattern in the less dense breast group than in the denser breast group, suggesting that MBDs in the less dense breasts tend to be clustered together. Moreover, we also observed that MBDs in the mature women (<65 years old) tended to be clustered in the middle zone, while in older women (>64 years old) they tended to be clustered in both the posterior and middle zones. (4) Conclusions: There is an inverse relationship between the amount of MBD (overall PD in the breast) and age, and a different clustering pattern of MBDs between the older and mature women.