Deep Learning of Position-Aligned Dynamic Cell Deformation Trajectories for Ovarian Cell Phenotyping in Hyperbolic Microchannels

Abstract Cell deformation during microfluidic transport evolves continuously, whereas cell phenotyping commonly relies on geometric descriptors measured at one or a few predefined channel positions. Such fixed-position measurements capture selected deformation states but overlook how the cellular response progresses along the flow path. Combined with high-speed imaging, hyperbolic microchannels capture this progression as a continuous trajectory, enabling the shift from static morphological measurements to a dynamic response analysis. However, cell-to-cell differences in transit velocity cause equivalent image frames to represent different channel positions and hydrodynamic conditions, preventing a direct comparison of frame-indexed deformation trajectories. Here, we develop a trajectory-resolved framework that converts high-speed microfluidic image sequences into position-aligned dynamic cell deformation trajectories and uses deep learning to model their cell line-specific evolution for label-free ovarian cell phenotyping. Geometric and kinematic trajectories were extracted from three ovarian cancer cell lines (A2780, OVCAR-3, and SKOV-3) and one nonmalignant ovarian epithelial cell line (IOSE-80) at three flow rates. The trajectories were reparameterized by the axial position and resampled on a common spatial grid. Class-specific one-dimensional convolutional neural networks predicted downstream trajectories from the inlet state and flow conditions. An unknown cell was classified by comparing its measured trajectory with class-specific predictions, and prediction errors in the width, area, perimeter, and axial velocity were integrated by soft voting. The framework achieved 93.26% accuracy, 94.07% macro precision, 90.25% macro recall, and a 91.85% macro-F1 score on the held-out test data. These results support extending microfluidic cell phenotyping from isolated geometric states to the position-aligned analysis of dynamic deformation trajectories.