Figure 5 | Scientific Reports

Figure 5

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

Figure 5

Comparison of StarDist and pixel-based model nucleus segmentation in tissue regions with high cell density or high autofluorescence. (A) Representative validation field-of-view of immunofluorescence microscopy image of foreskin tissue stained with DAPI (nuclei), showing examples of image crops (70 × 70 µm) with either (aF) high autofluorescence or (hD) high cell density (imaged at 200 × total magnification). Subsequent panels show the same FOV/image crops with nuclei segmentation by (B) manual tracing (gold standard), (C) the pixel-based model, and (D) the StarDist model. Yellow arrows point to areas with autofluorescent collagen fibers and cyan arrows point to areas with tightly packed cells.

Back to article page