Fig. 8: Data pipeline for Universal Immunohistochemistry (UIHC) artificial intelligence (AI) model. | npj Precision Oncology

Fig. 8: Data pipeline for Universal Immunohistochemistry (UIHC) artificial intelligence (AI) model.

From: A universal immunohistochemistry analyzer for generalizing AI-driven assessment of immunohistochemistry across immunostains and cancer types

Fig. 8: Data pipeline for Universal Immunohistochemistry (UIHC) artificial intelligence (AI) model.The alternative text for this image may have been generated using AI.

a Example of annotation process; patches were extracted from whole slide images (WSIs), then cells were manually annotated by expert pathologists. WSIs were split into 0.04 mm2 patches (resized to 1024 × 1024 pixels at 0.19 microns-per pixel). b Patch-level annotation counted by its positivity (negatively stained Tumor Cell [TC−] or positively stained Tumor Cell [TC+]). c The number of WSI in the WSI-level dataset only for testing.

Back to article page