Related Experiment Video
Updated: Feb 7, 2026

Studying Triple Negative Breast Cancer Using Orthotopic Breast Cancer Model
Published on: March 20, 2020
TILseg: Automated Whole Slide-Level Spatial Scoring of Tumor-Infiltrating Lymphocytes Reveals Prognostic Patterns in
Lisa L Carr1, Abishek Sankaranarayanan1, Khanh Ha1
1Department of Chemical Engineering, University of Washington, Seattle, WA 98195, USA.
An AI tool for analyzing stromal tumor-infiltrating lymphocytes (sTILs) in triple-negative breast cancer (TNBC) offers reproducible and accurate prognostication. This digital pathology approach improves patient risk stratification for better therapeutic outcomes.
Area of Science:
- Computational pathology
- Biomarker discovery
- Breast cancer research
Background:
- Stromal tumor-infiltrating lymphocytes (sTILs) are key biomarkers for predicting treatment response and survival in triple-negative breast cancer (TNBC).
- Current manual sTIL evaluation by pathologists lacks reproducibility due to inter-reader variability.
- There is a need for objective and standardized methods for sTIL assessment in TNBC.
Purpose of the Study:
- To develop and validate an artificial intelligence (AI)-driven pipeline for automated segmentation and scoring of sTILs in whole slide images of TNBC.
- To assess the prognostic value of AI-derived global and spatially resolved sTIL scores compared to manual scoring.
- To investigate the impact of sTIL spatial distribution relative to epithelial regions on patient outcomes.
Main Methods:
- Development of an AI pipeline (TILseg) for processing H&E-stained whole slide images to quantify sTILs.
- Validation of the pipeline on two independent TNBC patient cohorts (n=57 and n=43) with clinical outcome data.
- Comparison of AI-derived global and spatial sTIL scores with expert manual scoring and patient prognostication (response to neoadjuvant chemotherapy, recurrence-free survival).
Main Results:
- The AI-driven global TILseg scoring demonstrated strong concordance with expert pathologists (Spearman R = 0.84-0.89).
- Global TILseg scores provided significantly improved patient stratification compared to manual scoring (p=0.0191 vs p=0.0663).
- Spatially resolved TILseg, particularly TILs within 50 µm of epithelial regions, showed higher prognostic significance for recurrence-free survival and pathological complete response than global or manual scores.
Conclusions:
- An automated, spatially resolved AI tool for sTILs scoring enhances patient risk stratification.
- This AI approach improves prognostication based on treatment response and recurrence-free survival in TNBC.
- The developed AI tool establishes sTILs as a robust diagnostic and prognostic biomarker, particularly for TNBC patients.
Related Concept Videos
Introduction to z Scores
z scores...
Introduction to z Scores
z scores...
z Scores and Area Under the Curve
Cancer Stem Cells and Tumor Maintenance
Cancer stem cells are thought to originate from tissue-specific normal stem cells or progenitor cells. The normal stem cells usually reside in...
Scalar and Vector Triple Products
The scalar triple product is the dot product of a vector with the cross product of two vectors....
Design Example: Designing Water Slide
Bernoulli's principle determines the water's velocity along the slide....

