跳到正文
Cell Reports Physical Science· Jennings·收录于· 6 天前

Improving label-free cell classification through offline fusion of mechanical and molecular phenotypes

Improving label-free cell classification through offline fusion of mechanical and molecular phenotypes

原文摘要

出版方登记 · OpenAlex

Label-free approaches that capture both biochemical composition and physical state are needed to distinguish closely related cell populations without destructive assays. Here we present a multimodal screening framework that integrates two high-throughput, label-free techniques: broadband coherent anti-Stokes Raman scattering (BCARS) spectroscopy for molecular fingerprints and real-time deformability cytometry (RT-DC) for mechanical phenotyping. The RT-DC instrument incorporates a novel axicon-based darkfield illumination scheme that produces annular excitation to enhance optical contrast during microfluidic deformation while preserving measurement throughput. Mechanical features from RT-DC and broadband molecular signatures from BCARS were combined through offline data fusion to form a joint feature space. Across differentiated HL60/S4 myeloid lineages and breast cancer cells stratified by endothelial cell-selective adhesion molecule (ESAM) expression, integrating modalities improved discrimination relative to either alone, achieving >75% accuracy with linear models and ~90% with nonlinear classifiers. These results establish multimodal mechano-molecular screening as a scalable, label-free strategy for identifying closely related cellular phenotypes.

页面未能读取全文,以上为出版方登记的摘要;方法与图表见原文

来源:Cell Reports Physical Science · doi.org