Computational Staining of Pathology Images to Study the Tumor Microenvironment in Lung Cancer
Principal Investigator
Name
Liangrui Pan
Degrees
Ph.D
Institution
Hunan university
Position Title
student
Email
About this CDAS Project
Study
NLST
(Learn more about this study)
Project ID
NLST-1057
Initial CDAS Request Approval
May 9, 2023
Title
Computational Staining of Pathology Images to Study the Tumor Microenvironment in Lung Cancer
Summary
The spatial organization of different cell types in tumor tissue reveals important information about the tumor microenvironment (TME). To facilitate the study of cellular spatial organization and interactions, we developed Histology-Based Digital Staining, a deep learning-based computational model, from standard hematoxylin and eosin-stained pathology images of lung adenocarcinoma Segment the nuclei of tumors, stroma, lymphocytes, macrophages, nuclear stream, and red blood cells. Using this tool, we identified and classified nuclei and extracted 48 cellular spatial organization-related features that characterize the TME. Using these features, we developed a prognostic model from the National Lung Screening Trial dataset and independently validated the model in the Cancer Genome Atlas lung adenocarcinoma dataset, where the predicted high-risk group showed significantly worse survival It was lower than the low-risk group (P = 0.001), with an HR of 2.23 (1.37-3.65) after adjusting for clinical variables. Furthermore, image-derived TME features were significantly correlated with gene expression of biological pathways. For example, transcriptional activation of the T cell receptor and programmed cell death protein 1 pathways was positively correlated with lymphocyte density detected in tumor tissue, whereas expression of extracellular matrix tissue pathways was positively correlated with stromal cell density. In conclusion, we demonstrate that the spatial organization of different cell types predicts patient survival and correlates with gene expression of biological pathways.
Aims
1.Train a Deep Learning Model for Nuclei Classification and Segmentation
2.Using the model, count the number of different cells in the area to obtain quantified results
Collaborators
Liangrui Pan1, Liwen Xu1, Dazhen Liu1, Yutao Dou1, Lian Wang1, Mingting Liu1, Zhichao Feng2, Pengfei Rong2, Shaoliang Peng1
1College of Computer Science and Electronic Engineering, Hunan University, Changsha, China
2Department of Radiology, Third Xiangya Hospital, Central South University, Chang Sha, China