Related Experiment Video
Updated: Jul 7, 2025
11:00
Visualization, Quantification, and Mapping of Immune Cell Populations in the Tumor Microenvironment
Published on: March 25, 2020
17.2K
True-T - Improving T-cell response quantification with holistic artificial intelligence based prediction in
Yasmine Makhlouf1, Vivek Kumar Singh1, Stephanie Craig1
1Precision Medicine Centre of Excellence, Health Sciences Building, The Patrick G Johnston, Centre for Cancer Research, Queen's University Belfast, Belfast BT9 7AE, UK.
Computational and Structural Biotechnology Journal
|December 26, 2023
Summary
A new AI method, True-T, quantifies T-cells in colorectal cancer (CRC) using immunohistochemistry images. This tool provides prognostic information for stages II-IV CRC patients, improving upon existing standards.
Area of Science:
- Oncology
- Immunology
- Artificial Intelligence
- Computational Pathology
Background:
- T-cell lymphocytes in the tumor microenvironment are key to adaptive immunity in cancer.
- Quantifying T-cells in tumors is a potential diagnostic aid but not standard practice.
- Current methods lack efficiency and standardization for T-cell quantification in routine diagnostics.
Purpose of the Study:
- To develop and validate an AI-based method, True-T, for accurate T-cell quantification in colorectal cancer (CRC) using immunohistochemistry (IHC) images.
- To assess the prognostic value of True-T in predicting five-year survival rates for CRC patients.
- To evaluate the generalizability of the True-T algorithm across various solid tumor types.
Main Methods:
- Developed True-T, an AI pipeline using U-Net with ResNet-34 encoder for T-cell segmentation (CD3, CD4, CD8) in IHC images.
- Implemented density estimation from segmented masks to derive prognostic indicators.
- Validated the method on 1041 patients from four institutions and tested CD3 component on 13 solid tumors.
Main Results:
- True-T accurately segments T-cells (CD3, CD4, CD8) in IHC images, demonstrating generalization capabilities.
- T-cell density estimation provided valuable prognostic information for stages II-IV CRC patients.
- The CD3 component showed universal accuracy across 13 distinct solid tumors, indicating broad applicability.
Conclusions:
- True-T offers a robust, AI-driven solution for T-cell quantification in CRC, surpassing current quantitative standards.
- The method provides significant prognostic value, aiding in personalized treatment strategies for cancer patients.
- True-T demonstrates potential for widespread clinical application in oncology diagnostics and prognostics.
