• 2019-07
  • 2019-08
  • 2019-09
  • 2019-10
  • 2019-11
  • 2020-03
  • 2020-07
  • 2020-08
  • br RESULTS br Table shows the characteristics of our study


    Table 1 shows the characteristics of our study cohort and lesions. There was no statistically significant difference in age
    TABLE 1. Characteristics of the Study Cohort and Lesions
    Characteristics Triple-negative HER2-enriched Luminal (A + B)
    No. of patients
    HER2, human epidermal growth factor receptor 2.
    TABLE 2. Overall Classification Performance (Format: AUC/Accuracy) of the Three Binary Classifications of Breast Cancer Subtypes
    AUC, area under receiver operating characteristic curve; CC, craniocaudal; HER2, human epidermal growth factor receptor 2; MLO, medio-lateral oblique.
    (all P > .05) among the three subtype groups. The majority of the breast cancers are mass (49.5%) and nonmass (49.5%, including architectural distortion and Lipo3000 density) with only two microcalcifications. Of the 331 cancer cases, 305 (92.1%) were diagnosed as invasive ductal carcinomas of no special type and 26 (7.9%) were diagnosed as invasive car-cinomas of special type (including 10 mucous carcinomas, 9 lobular carcinomas, and 7 papillary carcinomas).
    Table 2 shows the comparisons of the overall classification performance. As shown, AUC ranged from 0.695 to 0.865, whereas accuracy ranged from 0.634 to 0.796 in all the experiments. In terms of AUC, MLO view shows a higher AUC than the CC view across the three binary classifications, and when the CC and MLO view were combined, AUC increased except for the luminal vs nonluminal classification. In terms of accuracy, there were no consistent trends between the CC and the MLO view data across the three subtype clas-sifications; however, accuracy of the combination of the CC and MLO view outperformed either of them alone. In gen-eral, it is clear that the combination of CC and MLO view yielded the overall best classification performance.
    In terms of the most significant features selected by the LASSO method, there were 11, 10, and 12 top-ranked fea-tures for the CC view, MLO view, and their combination, respectively. Figure 1 shows the boxplots of the 12 selected features from CC or MLO view images when they were combined for classification, including perimeter (CC), con-cavity (CC), correlation (CC), inverse different moment (CC), roundness (MLO), concavity (MLO), 10th Fourier coefficients (MLO), 24th Fourier coefficients (MLO), gray 
    mean (MLO), correlation (MLO), energy (MLO), and inverse different moment (MLO). Note that the three fea-tures, concavity, correlation, and inverse different moment, in both CC and MLO views were selected. Of the 12 fea-tures, 4 showed a statistically significant (P < .05) difference: concavity (P = .027), correlation (P = .0015), roundness (P = .00016), and gray mean (P = .026). More specifically, Polycistronic mRNA can be seen that the concavity values of triple-negative sam-ples tend to be smaller than those of HER2-enhanced and luminal samples (Fig 1f and b), the correlation values of lumi-nal lesions tend to be larger than those of the other lesion types (Fig 1c), the roundness values of triple-negative samples were significantly larger than those of HER2-enahnced and luminal lesions, and Figure 1h shows that the gray mean val-ues of triple-negative samples tend to be larger than those of the other subtypes.
    In this study, we employed a radiomic approach to investigate the potential association between breast cancer molecular subtypes and quantitative imaging features extracted from digital mammogram images. Our results on the three binary classifications of subtypes (ie, triple-negative vs other types, HER2-enhanced vs other types, and luminal vs other types) showed that a set of such quantitative radiomic features is pre-dictive of the molecular subtypes of breast cancer.
    Although several previous studies have shown that breast MRI-derived radiomic features are associated with the sub-types (11,13,24 26), our study complements answering a
    Academic Radiology, Vol 26, No 2, February 2019 MAMMOGRAPHIC RADIOMIC FEATURES
    question of whether imaging features in digital mammography would have a similar association effect or not. Although mam-mography is different from MRI in terms of potential imaging traits that they may be able to capture, our study demonstrated the value of mammogram images that capture only morpho-logic or anatomic properties of the breast in helping assess breast cancer subtypes from the image-derived features.
    Several previous studies have shown some correlation between breast cancer subtypes and certain qualitative mam-mographic characteristics or reading experience. For exam-ple, triple-negative subtypes may be more likely to manifest as an ill-defined mass, whereas non triple-negative subtypes more likely to present as a spiculated mass; HER2-enriched subtypes are often characterized by structural distortion, and luminal lesions are more often to present like oval on mam-mogram images. Our finding is in line with these previous studies and we showed a further relationship between the molecular subtypes and the quantitative radiomic imaging features automatically extracted from digital mammograms.