Mechanistic insights into cellular deformation enable enhanced extensional-flow cytometry for label-free classification and sorting

Cellular biomechanics have emerged as powerful, label-free indicators for stratifying heterogeneous cell populations across disease contexts. However, existing deformability-based cytometry techniques often suffer from limited specificity, low interpretability, and poor compatibility with real-time sorting, limiting their translational utility in functional screening. Here, we demonstrate an extensional-flow cytometry platform for tumor cell profiling and sorting, enabled by a mechanistic reinterpretation of cellular deformation dynamics. Fluid–structure interaction simulations facilitated a mechanistic reinterpretation of strain-induced morphological transitions, allowing deformation phenotypes to be reliably captured and interpreted through image-derived features. Then we developed a novel lightweight detection algorithm incorporating auto-localization filters and a normalized block attention module to enhance spatial precision and morphological sensitivity, and achieved a mean average precision upon 96.8%. The final sorting is accomplished through a fully integrated pipeline comprising droplet encapsulation, electrostatic charging, and voltage-controlled deflection, yielding a sorting purity of 90.2% ± 4.4% while maintaining cell viability above 95%, which approaches the state-of-the-art technology for image-driven, label-free deformability-based systems. Our work establishes a robust and scalable cytometry platform that bridges mechanistic insight with real-time, label-free sorting, offering an interpretable solution for mechanophenotyping and functional cytometric decision-making.