3D-printed real-time biosensing system integrating deep learning for label-free in vitro T cell culture analysis

TY - JOUR

T1 - 3D-printed real-time biosensing system integrating deep learning for label-free in vitro T cell culture analysis

AU - Li, Bo

AU - Guo, Xiaoliang

AU - Yao, Jialin

N1 - Publisher Copyright: Copyright © 2026. Published by Elsevier B.V.

PY - 2026/10/1

Y1 - 2026/10/1

N2 - Recent advances in T cell-based immunotherapies highlight the urgent need for precise and dynamic monitoring across the entire cell culture pipeline. In contrast to conventional morphological assessment methods remain limited by subjectivity and static analytical paradigms, rendering them insufficient for real-time, in-process monitoring in T cell manufacturing. This capability is particularly critical for emerging metabolic reprogramming strategies, where optimizing therapeutic outcomes requires real-time tracking of dynamic cellular responses that static methods cannot provide. Here, we report a fully automated, label-free monitoring platform constructed with 3D-printed modular components that integrates automated T cell culture, real-time bright-field imaging, and deep learning algorithms. This system enables continuous tracking of T cell morphology, viability, and migratory behavior, achieving a mean average precision above 96% in cell detection and phenotypic characterization. This cost-effective architecture supports data-driven optimization of T cell expansion and provides a versatile process analytical technology for cancer immunotherapy research.

AB - Recent advances in T cell-based immunotherapies highlight the urgent need for precise and dynamic monitoring across the entire cell culture pipeline. In contrast to conventional morphological assessment methods remain limited by subjectivity and static analytical paradigms, rendering them insufficient for real-time, in-process monitoring in T cell manufacturing. This capability is particularly critical for emerging metabolic reprogramming strategies, where optimizing therapeutic outcomes requires real-time tracking of dynamic cellular responses that static methods cannot provide. Here, we report a fully automated, label-free monitoring platform constructed with 3D-printed modular components that integrates automated T cell culture, real-time bright-field imaging, and deep learning algorithms. This system enables continuous tracking of T cell morphology, viability, and migratory behavior, achieving a mean average precision above 96% in cell detection and phenotypic characterization. This cost-effective architecture supports data-driven optimization of T cell expansion and provides a versatile process analytical technology for cancer immunotherapy research.

KW - Deep learning

KW - Label-free sensing

KW - Process analytical technology

KW - T cell culture

UR - https://www.scopus.com/pages/publications/105036556891

U2 - 10.1016/j.talanta.2026.129892

DO - 10.1016/j.talanta.2026.129892

M3 - Article

AN - SCOPUS:105036556891

SN - 0039-9140

VL - 308

JO - Talanta

JF - Talanta

M1 - 129892

ER -