Description
Cell viability is a fundamental parameter in biomedical research, as it reflects the functional state of a cell population by integrating key aspects of cellular health, including membrane integrity, metabolic activity, and proliferative capacity [Khalef et al.(2024)]. It plays a critical role in diagnostics and pharmacological studies, supporting the investigation of pathological processes, therapeutic efficacy, and host–pathogen interactions [Archambaud et al.(2024)].
Conventional viability assays are predominantly label-based, relying on chemical markers. Although accurate, they are invasive, often limited to end-point measurements, require complex instrumentation, and are ill-suited for integration into portable platforms [Madorran et al.(2025)]. These limitations motivate the development of non-invasive, on-chip systems able of evaluating cell viability while preserving sample integrity and reducing operational complexity.
This work presents a novel mechanobiological approach for label-free, on-chip cell viability assessment based on hydrodynamic manipulation in microfluidic channels. The proposed paradigm exploits the hypothesis that cells respond to hydrodynamic mechanical stresses according to their biological condition. In more details, cells are subjected to oscillatory hydrodynamic forcing, which, maintaining them out of equilibrium with respect to the surrounding fluid, generates characteristic motion signatures reflecting key biophysical properties, including cell size, density, deformability, and membrane integrity [IT Patent n.102025000005856].
In this context, an automated, label-free, image-based system was developed to extract these hydrodynamic signatures using a Digital Particle Image Velocimetry (DPIV)-based algorithm [Torrisi et al.(2023)]. Initial assessments successfully differentiated viable and apoptosis-induced yeast cells from inert particles based on their hydrodynamic response [Cutuli et al.(2025), Biomed. Signal Process. Control]. While inert particles passively follow the imposed flow, viable cells exhibit stronger mechanical opposition, whereas apoptotic cells display intermediate behaviour consistent with partial loss of mechanical integrity. Further validation across different cell lines, including HL60 cells, demonstrated sensitivity to unhealthy states induced by exposure to 5-azacytidine for 24h and 48h. Longer treatment times resulted in progressively weaker opposition to flow, consistent with gradual degradation of cellular biophysical properties. These results were benchmarked against standard biochemical viability assays (MTT), confirming the reliability of the proposed approach.
Despite its accuracy, the DPIV-based analysis is computationally intensive, thus limiting real-time applicability. To overcome this limitation, a lightweight deep learning model (APVnet) was developed to extract cell hydrodynamic signatures in real-time, achieving millisecond-scale inference with a minimal memory footprint [Cutuli et al.(2025), under review in Eng. Appl. Artif. Intell.]. Finally, to enhance portability, a smartphone-based reflective microscope was designed and prototyped to adapt the methodology to point-of-care scenarios and resource-limited environments, where access to standard, bulky and expensive, optical microscopes may be constrained. The system leverages the smartphone’s built-in flash and a reflective module to enable bright-field imaging using native components. Benchmarking against a conventional microscope showed no statistically significant differences in estimating cellular dynamic responses (p-value∈[0.4553,0.5597]), supporting its suitability for on-chip viability assessment outside standard laboratory settings.
Overall, the proposed platform combining hydrodynamic manipulation, label-free optical monitoring, AI-enhanced real-time analysis in either lab-microscopes and smartphone-based solution, represents a significant step toward automated cell viability assessment in microfluidic environments, with strong potential for next-generation point-of-care applications.