Table 3.
Confusion matrix showing machine learning classification of immune cell phenotypes.
| M1 Macrophage | M2 Macrophage | Naïve Macrophage | Monocyte (Day 0) | Monocyte (Day 6) | ||
|---|---|---|---|---|---|---|
| Cell Type | M1 Macrophage | 92.4 | 0.4 | 4.1 | 1.1 | 2.0 |
| M2 Macrophage | 6.3 | 89.0 | 4.3 | 0.0 | 0.3 | |
| Naïve Macrophage | 3.0 | 2.8 | 88.8 | 0.4 | 5.1 | |
| Monocyte (Day 0) | 7.4 | 0.5 | 2.1 | 90.0 | 0.0 | |
| Monocyte (Day 6) | 10 | 6.8 | 26.8 | 1.8 | 54.6 |
Data is presented for all cells as a percentage of correctly classified cells. Approximately 500 observations were classified for each cell type across 152 images using a random forest classifier. The number of correctly classified cells is presented as a percentage, with true positives highlighted on the diagonal in red. Confusion matrices for all 5 classifier types used in this work are included in the supplementary section.