Table 1. The average of the three evaluation metrics obtained for the 19 images using the Pix2Pix model and the cycle CGAN model are given for training and testing dataset. For reference purpose, the three metrics were also calculated with the same histopathologically stained H&E image. It is seen that MSE values are very large for both models, whereas SSIM and CSS are almost similar for both models. This means that the pixel values of computationally stained H&E images are different, but the overall structural and color information is acceptable. Furthermore, the metric values for training and testing dataset do not have a large difference, which indicates that the models are minimally overfitted.

	MSE	SSIM	CSS
	Training dataset
Pathological stained H&E image	0.00	1.00	1.00
Pix2Pix stained H&E image	4.69$\times {10}^3$	0.52	0.93
Cycle CGAN stained H&E image	10.26$\times {10}^3$	0.49	0.91
	Testing dataset
Pathological stained H&E image	0.00	1.00	1.00
Pix2Pix stained H&E image	4.27$\times {10}^3$	0.60	0.94
Cycle CGAN stained H&E image	7.79$\times {10}^3$	0.59	0.93