The Pearson correlation coefficients of the
The Pearson correlation coefficients of the feature of irreg-ularity were 0.43, 0.71, 0.59, and 0.80 with the features of lesion volume, effective diameter, surface area, and maximum linear size, respectively.
Feature selections for the scenarios of using all features and all features except those relating to size demonstrate the importance of the lesion shape (irregularity) and the enhance-ment texture features (Fig 3).
Academic Radiology, Vol 26, No 2, February 2019 RADIOMICS IN DIAGNOSING LUMINAL A CANCERS
Figure 3. Use of features for each fold, with and without the features of size available to the linear discriminant analysis classifier algorithm for feature selection. The number in each square indicates for a given fold how many features from the associated category were selected for clas-sification. The features of irregularity and 13(S)-HODE were chosen in all folds, in each feature selection scenario.
TABLE 2. Difference in Area Under the Receiver Operating Characteristic Curve and 95% Confidence Interval for Comparisons of Classification Protocols
Classification Protocol Comparison DAUC Interval of the Difference P Value Conclusion
Signature based on feature selection on all features excluding 0.005 Superior
Signature based on feature selection on all features excluding 0.3 Equivalent
size compared to that on all features
AUC, area under the receiver operating characteristic curve.
P values shown are after correction for multiple comparisons.
Statistical conclusion was based upon superiority and equivalence testing.
The AUCs for each classification protocol demonstrate the performance of the three classification protocols in the clini-cal task of distinguishing between benign lesions and luminal A breast cancers (Fig 4).
The AUC for maximum linear size alone was significantly different from the AUCs for both feature selections using all features (P = .005) and without size features included (P = 0.005). However, the AUC curves for both feature selections with and without size features included failed to show a significant difference from each other (P = 0.3). The AUC values do not approach nonequivalence until the statis-tical equivalence margin is less than 2% (Table 2).
The large number of actual lesions in the present study (over
500) and the emphasis of comparing benign lesions to a spe-cific breast cancer subtype (luminal A) offer a focused investi-gation into radiomic features useful for the clinical task of malignancy classification, as a majority of breast cancers are of luminal A type. The segmentation of lesions by an algorithm that required only a single seed point per lesion contributed objectivity to the feature extractions, whereas the use of LDA allowed for consideration of multiple features for the classifi-cation task, resulting in a lesion signature.
Figure 4. Receiver operating characteristic curves for the three different classification protocols from the 10-fold cross-validation, using max-imum linear size alone, radiomics lesion signature obtained through nested feature selection, and radiomics lesion signature obtained through nested feature selection excluding size features, in the task of distinguishing between benign lesions and luminal A cancers. The legend gives the AUC for each with the 95% confidence interval in brackets. AUC, area under the receiver operating characteristic curve.
A previous study reported that luminal A cancers were strongly associated with the Breast Imaging Reporting and Data System (BI-RADS) (30) descriptors of “irregular shape,” “spiculation,” “irregular margin,” “rim enhancement,” and “dark internal septation” (31). Using stepwise feature selec-tion and LDA demonstrated that the radiomic feature of irregularity, which describes the so-called roughness of a lesion (20), played a prominent role in the classification of benign lesions and luminal A breast cancers, regardless of whether or not features related to size were included in fea-ture selection. Notably, classification using the radiomic fea-ture of irregularity alone resulted in an AUC of 0.668 (0.626, 0.709) and failed to show a significant difference in perfor-mance compared to using maximum linear size alone (P = 0.19). Despite this and the substantial correlation of the irregularity feature to size features (correlation coefficients between 0.42 and 0.8), feature selection demonstrated the additive benefit of using irregularity and a few other features related to shape, morphology, and texture over using only size in classification. This work suggests that the degree of roughness is particularly but not uniquely important in the classification of lesions as benign lesions or luminal A cancers. Texture features describe spatial variation in the signal