Table 7.
Task and performance comparisons between the presented study and previous studies on breast cancer detection.
| Reference (Year) |
Task | Dataset Used | Ave_Sen (%) |
Ave_Spe (%) |
Acc (%) |
AUC |
|---|---|---|---|---|---|---|
| This study (2022) |
Classification of BI-RADS 0, 1, 2, 3, 4A, 4B, 4C, 5 (8 categories) |
Private (1490 cases, 5733 images) | 95.31 | 99.15 | 94.22 | 0.972 |
| [8] (2021) |
Malignancy prediction of BI-RADS 4 micro-calcifications (2 classes) |
Private (384 cases, 824 images) | 85.3 | 91.9 | - | 0.910 |
| [11] (2021) |
Mass malignancy classification (2 classes) |
DDSM (2578 cases, 10,312 images) | 89.8 @ 2 FPPI 1 | - | - | - |
| Private (2807 cases, 11,228 images) | 96.2 @ 2 FPPI | - | - | - | ||
| [13] (2020) |
BI-RADS 2-5 classification for breast masses (4 categories) |
DDSM (500 images) |
84.5 | 94.25 | 84.5 | - |
| [22] (2019) |
Normal, benign calcification, benign mass, malignant calcification, malignant mass (5 classes) |
DDSM + CBIS-DDSM (2339 images) |
- | - | 91 | 0.98 |
1 FPPI: false positive per image.