Figure 4 | Scientific Reports

Figure 4

From: Comparison between a deep-learning and a pixel-based approach for the automated quantification of HIV target cells in foreskin tissue

Figure 4

Automated cell segmentation using a pixel-based model systematically overcounts cells in the epidermis, and undercounts cells in the dermis. Cell counting was completed on 10 validation field-of-view images (600 × 600 µm) randomly selected from a set of 232 full section scans of foreskin tissues (inner and outer) stained for CD3, CD4, CCR5, and nuclei. Cell counting was completed on the full field-of-view images, and separately on the epidermis and dermis. The difference between automated counts and manual counts in (A) epidermis, (B) dermis, and (C) full tissue is displayed the percent difference from manual counts. Red dashed lines mark ± 10.0% difference from manual counts.

Back to article page