Table 3.

Quantitative comparison of infected tissue segmentation outcomes based on our pipeline and three recently published structures. The evaluations are based on recall, precision, and F score.

Approach	Precision (%)	Recall (%)	F score
DenseNet201 [1]	86%	89%	87%
Weakly supervised deep learning [71]	88%	90%	89%
Weakly supervised framework [4]	91%	89%	90%
Proposed CNN	88%	89%	88%
Proposed CNN+LDN	93%	91%	92%
Proposed CNN+fuzzy c-means	92%	94%	93%
Proposed CNN+fuzzy c-means+LDN	96%	97%	97%