Table 3 Benign vs Malignant microcalcifications: accuracy (Acc) and Area Under the Curve (AUC) performances
From: A machine learning approach on multiscale texture analysis for breast microcalcification diagnosis
Method | No. ROIs | Feature type | Classifier | Acc (%) | AUC (%) |
|---|---|---|---|---|---|
Chen et al. (2015) [17] | 300 | topological features | kNN | 85 | 91 |
Ren et al. (2012) [19] | 295 | statistical features | kNN | 82 | 86 |
Khehra et al. (2013) [21] | 380 | statistical, shape and | LS-SVM | 89 | 89 |
textural features | |||||
Strange et al. (2014) [25] | 300 | mereotopological features | Barcodes | 80 | 82 |
Hu et al. (2017) [38] | 150 | textural features | ELM | - | 92 |
Proposed approach | 260 | textural features and | RF | 88 | 92 |
#interest points/corners |