Skip to main content
. 2022 Apr 29;13:859323. doi: 10.3389/fimmu.2022.859323

Figure 1.

Figure 1

The workflow of overall study. Firstly, we evaluated whether CD8 expression signature could represent the TIME profiles by using the RNA-seq data in TCGA cohort (n = 1145). Then, machine learning model was trained and validated using DPH cohort (n = 221) with 18F-FDG PET/CT radiomics-clinical features to predict CD8 expression status. The model was then applied to predict TIME phenotypes in TCIA cohort (n = 39). The right row is the radiomics workflow.