Table 7 Comparison of proposed work with benchmarks.
Study | Used model | Accuracy (%) | Precision | Recall | F1-score |
|---|---|---|---|---|---|
Two-stage nuclei segmentation | 91.67% | 83% | 100% | 90% | |
ResNet50 | 92% | - | 95% | - | |
EfficientNet | 75% | 75% | 75% | 75% | |
Advanced-Deep-CNNs | 88% | - | - | - | |
InceptionResNetV2 | 91% | 82% | 84% | 83% | |
Vision Transformer | 88.6% | 88% | 88% | 88% | |
InceptionV3 | 92% | - | - | - | |
U-Net and YOLO | 93% | 93% | 94% | - | |
Vision transformer | 88.6 | 90.1% | 87.4% | 88.7% | |
YOLOv5 + Swin transformer | 94 | 94% | 92% | 93% | |
ResNet50 + ProtoNet | 88.2–88.9 | - | - | - | |
Proposed work | MobNAS | 97% | 97% | 97% | 97% |
