Fig. 4: The contribution of Segment Anything (SAM) masks. | Communications Biology

Fig. 4: The contribution of Segment Anything (SAM) masks.

From: A deep learning framework deploying segment anything to detect pan-cancer mitotic figures from haematoxylin and eosin-stained slides

Fig. 4

a Illustration of the quality assurance process for the MIDOG++ human subset. b F1 scores of the classifier using only RGB images (RGB Classifier), the classifier using additional SAM masks (RGB-M0 Classifier), and the model using reviewed SAM masks (RGB-M1 Classifier). The scores are presented using violin plots with individual data points (n = 10 independent experiments).

Back to article page