< Back to Article

Automated reporting of cervical biopsies using artificial intelligence

Fig 16

Slide level confusion matrices for imbalanced cervical dataset using XGBoost (Patch size = 1024 × 1024 pixels).

(a) Training set confusion matrix (Accuracy = 90.98%) (b) Validation set confusion matrix (Accuracy = 75.09%, Malignant sensitivity = 94.87%).

doi: https://doi.org/10.1371/journal.pdig.0000381.g016