Table 3 Training and testing set for lung models.
From: Accurate recognition of colorectal cancer with semi-supervised deep learning on pathological images
Model | Class | Dataset-lung | ||
|---|---|---|---|---|
Training set | Testing set | |||
Labeled | Unlabeled | |||
Lung-5%-SSL | Adenocarcinoma | 250 | 3750 | 1000 |
Squamous cell carcinoma | 250 | 3750 | 1000 | |
Benign | 250 | 3750 | 1000 | |
Total | 750/5% | 11,250/75% | 3000/20% | |
Lung-20%-SSL | Adenocarcinoma | 1000 | 3000 | 1000 |
Squamous cell carcinoma | 1000 | 3000 | 1000 | |
Benign | 1000 | 3000 | 1000 | |
Total | 3000/20% | 9000/60% | 3000/20% | |
Lung-5%-SL | Adenocarcinoma | 250 | – | 1000 |
Squamous cell carcinoma | 250 | – | 1000 | |
Benign | 250 | – | 1000 | |
Total | 750/5% | – | 3000/20% | |
Lung-20%-SL | Adenocarcinoma | 1000 | – | 1000 |
Squamous cell carcinoma | 1000 | – | 1000 | |
Benign | 1000 | – | 1000 | |
Total | 3000/20% | – | 3000/20% | |
Lung-80%-SL | Adenocarcinoma | 4000 | – | 1000 |
Squamous cell carcinoma | 4000 | – | 1000 | |
Benign | 4000 | – | 1000 | |
Total | 12,000/80% | – | 3000/20% |
