Table 2 Training and testing set for CRC patch-level models.
From: Accurate recognition of colorectal cancer with semi-supervised deep learning on pathological images
Model | Class | Dataset-PATT | Dataset-PAT | ||
|---|---|---|---|---|---|
Training set | Testing set | ||||
Labeled | Unlabeleda | ||||
Model-5%-SSL | Cancer | 1645 | 21,390 | 9828 | 14,317 |
Non-cancer | 1505 | 19,560 | 8991 | 85,683 | |
Total | 3150/5%b | 40,950/65%c | 18,819/30%d | 100,000 | |
Model-10%-SSL | Cancer | 3290 | 19,745 | 9828 | 14,317 |
Non-cancer | 3010 | 18,055 | 8991 | 85,683 | |
Total | 6300/10% | 37,800/60%e | 18,819/30% | 100,000 | |
Model-5%-SL | Cancer | 1645 | – | 9828 | 14,317 |
Non-cancer | 1505 | – | 8991 | 85,683 | |
Total | 3150/5% | – | 18,819/30% | 100,000 | |
Model-10%-SL | Cancer | 3290 | – | 9828 | 14,317 |
Non-cancer | 3010 | – | 8991 | 85,683 | |
Total | 6300/10% | – | 18,819/30% | 100,000 | |
Model-70%-SL | Cancer | 23,035 | – | 9828 | 14,317 |
Non-cancer | 21,065 | – | 8991 | 85,683 | |
Total | 44,100/70%f | – | 18,819/30% | 100,000 |
