Table 22 Ablation study performance with FCE images.
From: Feature fusion context attention gate UNet for detection of polycystic ovary syndrome
Model | With FCE images (%) | |||
|---|---|---|---|---|
Accuracy | Precision | Recall | F1-score | |
Baseline U-Net | 79.32 | 79.8 | 78.9 | 79.3 |
U-Net without FFCM | 80.10 | 80.6 | 79.5 | 80.0 |
U-Net with FFCM | 81.56 | 82.0 | 81.0 | 81.4 |
U-Net with Default Attention Gate | 82.74 | 83.2 | 82.3 | 82.7 |
U-Net with Modified Attention Gate | 83.56 | 84.0 | 83.2 | 83.6 |
FCAU-Net without FFCM and Attention Gate | 84.60 | 85.1 | 84.0 | 84.5 |
FCAU-Net without FFCM and Default Attention Gate | 86.78 | 87.4 | 86.3 | 86.8 |
FCAU-Net without FFCM and Modified Attention Gate | 88.15 | 88.7 | 87.6 | 88.1 |
FCAU-Net with FFCM and without Default Attention Gate | 96.24 | 96.7 | 95.8 | 96.2 |
FCAU-Net with FFCM and without Modified Attention Gate | 97.12 | 97.6 | 96.8 | 97.2 |
Proposed FCAU-Net | 99.89 | 99.9 | 99.8 | 99.9 |
