Figure 3 | Scientific Reports

Figure 3

From: A machine learning approach toward automating spatial identification of LAG3+/CD3+ cells in ulcerative colitis

Figure 3

The semi-automated analysis process consisted of both an (a) assay pipeline and (b) analysis pipeline. (a) In the assay pipeline, FFPE tissue sections were stained using multiplex IF to detect CD3 and LAG3 with DAPI counter stain for nuclear detection. Subsequently, the coverslip was removed and the same sections were stained with H&E. Whole slide images of the IF and H&E stains were captured on a slide scanner capable of both fluorescence and brightfield imaging. (b) In the analysis pipeline, an AI/ML algorithm was implemented to segment the tissue sub-compartments based on the H&E image. The fluorescence image and brightfield images were co-registered using commercial software (Halo, Indica Labs). The same software was utilized to perform spatial cell analysis in annotated regions transferred from the AI/ML algorithm.

Back to article page