Table 6.
Algorithms for the detection of MAs.
| Authors | Method | Database(s) used | No. of images | Performance measure |
|---|---|---|---|---|
| Walter et al. [88] | Gaussian filtering, top-hat transformation | Non-public dataset | 94 | SE 0.885 (FPI 2.13) |
| Spencer et al. [89] | Morphological operators, matched filtering | Non-public dataset | NA | SE 0.824, SP 0.856 |
| Niemeijer et al. [91] | kNN pixel classification | Non-public dataset | 140 | SE 1.00, SP 0.87 |
| Mizutani et al. [92] | Double-ring filter, neural network classification | ROC | 50 | SE 0.648 (FPI 27.04) |
| Fleming et al. [93] | Contrast normalization, watershed region growing | Non-public dataset | 1441 | SE 0.854, SP 0.831 |
| Abdelazeem [94] | Circular Hough transformation | Non-public dataset | 3 | NA |
| Lázár and Hajdu [96] | Directional cross-section profiles | Non-public dataset; ROC | 110 | RS 0.233; RS 0.423 |
| Zhang et al. [97] | Multi-scale correlation coefficients | ROC | 50 | RS 0.357 |
| Antal and Hajdu [115] | Ensemble-based detection | ROC | 50 | RS 0.434 |