Table 2 ROC-AUC and log loss results for aggressive classification on the core needle biopsy test set using existing adenocarcinoma classification models
From: Inference of core needle biopsy whole slide images requiring definitive therapy for prostate cancer
Existing deep learning models | ROC-AUC | Log loss |
|---|---|---|
Stomach ADC, AD (x10, 512) | 0.768 [0.734 - 0.808] | 1.443 [1.286 - 1.563] |
Stomach SRCC (x10, 224) | 0.787 [0.747 - 0.823] | 0.858 [0.768 - 0.949] |
Stomach poorly ADC (x20, 224) | 0.806 [0.771 - 0.840] | 0.542 [0.516 - 0.568] |
Colon ADC, AD (x10, 512) | 0.568 [0.518 - 0.606] | 1.499 [1.371 - 1.665] |
Colon poorly ADC (x20, 512) | 0.889 [0.861 - 0.914] | 0.415 [0.378 - 0.457] |
Pancreas EUS-FNA ADC (x10, 224) | 0.739 [0.703 - 0.782] | 0.639 [0.596 - 0.677] |
Breast IDC, DCIS (x10, 224) | 0.748 [0.705 - 0.784] | 1.450 [1.333 - 1.569] |