Table 4 Performance metrics of the five features with area under the curve > 0.6 to differentiate low and high Ki-67 expression on digital breast tomosynthesis
From: Breast cancer Ki-67 expression prediction by digital breast tomosynthesis radiomics features
Feature | Feature class | Accuracy | Sensitivity | Specificity |
|---|---|---|---|---|
Sphericity | Shape | 0.656 | 0.645 | 0.673 |
Autocorrelation | Grey level co-occurrence matrix | 0.677 | 0.682 | 0.653 |
Interquartile range | First order | 0.653 | 0.661 | 0.631 |
Robust mean absolute deviation | First order | 0.676 | 0.689 | 0.631 |
Short-run high grey-level emphasis | Grey level run length matrix | 0.698 | 0.742 | 0.534 |