Integrative Spatial Analysis of H&E and IHC Images Identifies Prognostic Immune Subtypes correlated with Progression Free Survival in HPV-Related Oropharyngeal Squamous Cell Carcinoma
Many people with a certain type of throat cancer (HPV+ OPSCC) get better, but about 20% have the disease come back or spread. It's important to find signs that can tell which patients might face these issues. Deep learning is good at spotting immune cells in certain tissue slides. But, it's not great at understanding their varied roles in tumors. By combining two types of image techniques, we hope to better understand these cells and group patients more effectively.
- The Jupyter notebook named KM_plots_paper_submission.ipynb produces all the plots required for the paper.
- plot-cox.R produce Figure 4 in the manuscript.
- The registration codes for the H&E and IHC images can be foung in here
- The data folder contains processed data for all the patients in a csv file format.
- For the TIL, Stroma, Tumor predictions on a tile level on H&E, we used pretrained model from here
- For the grid level prediction and patient level predictions of Immune inflamed, Immune excluded and Immune desert are performed using Immune_pheno.ipynb
© This project is licensed under GPL - see the LICENSE file for details.
This study was funded internally at Mayo Clinic, Florida. We thank the patients and their families.
Reach out to the Hwang Lab.
