Figure 2 | Scientific Reports

Figure 2

From: Automated detection of the HER2 gene amplification status in Fluorescence in situ hybridization images for the diagnostics of cancer tissues

Figure 2

The predication of the two-stage deep learning-based detection system for the HER2 gene amplification stage is in par with pathologists. (A) Interrater agreement (Light’s Kappa κ) among a team of three pathologists and between the team of pathologists and the nucleus detector (ND) and signals detector (SD), respectively, for all images, for high quality images and for low quality images. (B) Nucleus class-specific accuracies between the team of pathologists and the nucleus and signal detector. (C) Confusion matrices for all images with respect to the classification of nuclei among the group of the three pathologists, between the three pathologists and the nucleus detector and between the three pathologists and the signal detector, respectively. Light’s Kappa κ and accuracies are shown above each matrix.

Back to article page