Skip to main content
Fig. 2 | Journal of Translational Medicine

Fig. 2

From: Artificial intelligence for quantifying immune infiltrates interacting with stroma in colorectal cancer

Fig. 2

Association of Deep-TSR score, Deep-TIL score, and Deep-immune score with stroma-CD3 density. A A second CNN model (CNN-IHC) was used for tissue-level segmentation of IHC-stained WSI. The tissue types of the segmentation are the same as Fig. 1A. STR was used as the region of interest for WSI, and all CD3 + T-cells were segmented and counted within WSI. Then, the stroma-CD3 density was calculated by using the number of all CD3+ T cells divided by the STR area. B-D Student t-test was also used to compare the difference in stroma-CD3 density between groups with different scores (such as Deep-immune score 4 vs. 3) in primary cohort. E–G Student t-test was used in validation cohort to compare the difference in stroma-CD3 density between groups with different scores. (nsP > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, Student’s t-test). HE, hematoxylin and eosin; IHC, immunohistochemistry; WSI, whole-slide image; CNN, convolutional neural network; STR, stroma; TUM, tumor epithelium; TSR, tumor-stroma ratio; TIL, tumor-infiltrating lymphocyte

Back to article page