Table 3. Results of nuclei detection algorithm in [2, 28] trained and tested for different stain deconvolution algorithms.

Values show the mean and standard deviation for each of the precision, recall, and F1 score measures. Note that the evaluated algorithms are dynamically estimating stain colour based on current information. Thus, consistency of the algorithm could improve the detection accuracy. However, we did not include stain normalization in this experiments to avoid affecting the deconvolution results.

	Precision	Recall	F1 score
Proposed	0.809 ± 0.1972	0.419±0.2404	0.520±0.223
BCD [13]	0.469±0.384	0.399±0.419	0.352±0.363
Macenko et al. [11]	0.407±0.281	0.553 ±0.335	0.374±.0270
ICA [12]	0.370±0.351	0.416±0.262	0.288±0.240