Related Experiment Video
Updated: Apr 5, 2026
10:56
Endothelial Cell Transcytosis Assay as an In Vitro Model to Evaluate Inner Blood-Retinal Barrier Permeability
Published on: June 7, 2022
6.5K
A deep learning approach to assess transendothelial cell trafficking performance.
Thomas Michael Schumacher1, Elisabeth Marie Gottloeber1, Eric Koziel1
1Department of Internal Medicine I, Ulm University Hospital, 89081, Ulm, Germany.
Scientific Reports
|April 3, 2026
Summary
This study introduces an AI-powered live-cell imaging tool for analyzing transendothelial migration (TEM). The novel approach accurately quantifies T cell transmigration, offering a standardized method for disease research.
Area of Science:
- Immunology
- Biotechnology
- Computational Biology
Background:
- Transendothelial migration (TEM) is crucial for immune responses and disease progression, but conventional assays lack in vivo accuracy and comprehensive visualization.
- Existing flow-based adhesion assays offer advantages but are limited by manual analysis, operator bias, and poor scalability.
- Accurate assessment of TEM is vital for understanding diseases like autoimmune disorders and cancer and developing new treatments.
Purpose of the Study:
- To develop and validate a standardized, AI-driven live-cell imaging tool for analyzing transendothelial migration (TEM).
- To overcome the limitations of conventional TEM assays, including mimicking in vivo conditions and enabling scalable, objective analysis.
- To create a versatile platform for investigating TEM in various disease models.
Main Methods:
- Combined flow-based adhesion assays with AI-based analysis using a Keras/TensorFlow deep learning model for classifying cell transmigration phases.
- Trained the deep learning model on T cells from healthy donors and pancreatic cancer patients.
- Utilized live-cell imaging for real-time, high-resolution observation of the transmigration process.
Main Results:
- The AI model achieved a high accuracy of 91.6% in identifying and categorizing cell transmigration, exceeding the 80% threshold for reliable performance.
- The developed tool provides a fast, standardized, and scalable method for analyzing TEM, overcoming limitations of manual analysis.
- The AI model's architecture is adaptable for investigating TEM in diverse disease models.
Conclusions:
- An AI-powered live-cell imaging tool has been established, integrating flow-based assays with deep learning for accurate TEM analysis.
- This approach offers a powerful, standardized, and scalable solution for TEM research across various disease contexts.
- The tool enhances the potential for leveraging TEM insights in developing novel therapeutic strategies for immune-related diseases and cancer.
