Fig. 1.

ROI and FOV examples at 20× magnification.

1A. Within this FOV, each box designates an ROI. The blue border of each box indicates that the bounded ROI is ground truth “benign”.

1B. Within this FOV, each red box indicates an ROI with a ground truth of “cancer”. A green circle within an ROI represents an area where the model’s classification diverged from the ground truth. Thus, an ROI with a green circle inscribed in a blue box is a false positive, while an ROI with a green circle inscribed in a red box is a false negative. This FOV contains a significant amount of false-positive ROIs.