Skip to main content
Fig. 3 | Diagnostic Pathology

Fig. 3

From: Normalization of HE-stained histological images using cycle consistent generative adversarial networks

Fig. 3

Results gallery from our experiments on the Mitos-Atypia-14 challenge data set. Columns A-C refer to the image tiles scanned by the Aperio scanner (\(X_{A}^{real}\)) being mapped by the generator GB to produce the corresponding image in the domain of the Hamamatsu scanner (\(X_{B}^{fake}\)) and the reconstruction from mapping the image back to its source domain (\(X_{A}^{rec}\)), i.e \(X_{A}^{real} \xrightarrow {G_{B}} X_{B}^{fake} \xrightarrow {G_{A}} X_{A}^{rec} \). The same process is done in the reverse direction for image tiles scanned in domain B, i.e \(X_{B}^{real} \xrightarrow {G_{A}} X_{A}^{fake} \xrightarrow {G_{B}} X_{B}^{rec} \) (column D-F). Each row 1-4 presents another example tissue section

Back to article page