Single-cell vibration analysis for potential diagnostic applications.

Abstract

Vibrational frequency profiling (VFP) using optical tweezers has emerged as a promising technique to characterize single-cell behavior for diagnostic applications. These applications include cancer screening, identifying bacterial resistance to antibiotics, and assessing cellular response to metabolic treatments. Previously, we demonstrated that vibrational profiling can successfully differentiate samples in real time using single-cell vibrational signatures. However, the sensitivity of this approach and its ability to be applied across variable experimental conditions and cell lines have not been well identified. This study builds on previous work, demonstrating the feasibility of our established vibrational analysis to quantitively assess whether changes in experimental design can improve model separation of different cell types. U251 human glioblastoma cells and A549 human lung carcinoma cells were used as our model system. We illustrate this capability through three examples: 1) comparing an open Petri dish to closed microfluidic chambers, 2) synchronizing cells in the cell cycle, and 3) altering medium viscosity. These factors were selected for their potential to influence signal quality and model accuracy. Vibrations were collected using optical tweezers, were Fourier transformed to produce power spectra, and were then parameterized using peak detection, extracting area under the curve values. Peaks were aggregated across samples and classified using partial least squares discriminant analysis. Partial least squares discriminant analysis classification was performed with 70/30 train-test splits of data and 10-fold cross-validation. Petri-dishes yielded a more well-classified sample than microfluidics chambers (F1 score 0.89 vs. 0.79). Synchronized cells reduced vibrational variability and improved classification (F1 score 0.83 vs. 0.79). Increased viscosity produced slight classifier improvements (F1 score 0.81 vs. 0.79). These findings demonstrate that VFPs are sensitive to the mechanical and environmental context of cell measurement, highlighting that careful standardization of experimental conditions is crucial for developing reproducible and clinically translatable VFP-based diagnostic platforms.