RETRACTED ARTICLE: Mass classification method in mammograms using correlated association rule mining

Fully automated scheme for computer‐aided detection and breast cancer diagnosis using digitised mammograms

“…Table 6 shows that the proposed CAD system is superior over other approaches in reducing the number of false negatives, and consequently achieving the highest SN of 98.63%. Also, it can be noted that the SP of the proposed system is higher than the CAD systems of [39, 40, 43, 44], and is comparable with the CAD systems of [37, 38, 42]. The comparison illustrated in Table 6 reveals the superior performance of the proposed CAD system over other existing systems.…”

“…Note that the Q‐classifier can identify accurately samples from the minority class (patients have breast cancer) when dealing with imbalance datasets, and consequently, it provides superior SN results comparing to all other binary classifiers. This can be noted in Table 6, where the proposed system achieves the highest SN of 98.63% compared to all other CAD systems under comparison [37–44]. To the authors’ knowledge, the Q‐classifier is the first similarity‐based classifier that is designed for imbalance breast cancer data classification.…”

“…The proposed CAD system is compared with other recent CAD systems [37–44], and the results are shown in Table 6. It can be noted that we examined a larger number of mammograms than all other CAD systems under comparison.…”

Analysis of 2D singularities for mammographic mass classification

“…Though a number of shape- [5][6][7] and margin- [8][9][10] based features have already been proposed for this purpose, their performance highly depends on the segmentation of masses. On the other hand, features based on textural patterns of mammograms are widely used to distinguish between benign and malignant masses [11][12][13][14][15]. Liu et al [16] computed Haralick's features from Gray level co-occurrence matrix (GLCM), derived from the region around the contour of the mass, and achieved an area under the receiver operating characteristic (ROC) curve (A z ) of 0.70 for benign and malignant classification, respectively, with 309 DDSM images.…”