Fig. 6: Including mitotic-like figures (MLFs) and non-mitotic objects for training improves the detection. | Communications Biology

Fig. 6: Including mitotic-like figures (MLFs) and non-mitotic objects for training improves the detection.

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

Fig. 6

a Example of patches containing a mitotic figure (MF) (left) and an MLF (right). The MFs and MLFs are masked in green (Original data). The surrounding cellular components segmented by Segment Anything (SAM) are marked in light blue and are added to the MFs and MLFs (SAM-Aug data). b The precision, recall and F1 scores of the model trained with the Original data and the model trained with SAM-Aug data. The scores are presented using violin plots with individual data points (n = 5 independent experiments).

Back to article page