This doctoral work presents the integration of a quantitative phase imaging (QPI) digital holographic microscope (DHM) with a multilayer soft lithography stop-and-go microfluidic system for the high-contrast, label-free phenotyping of single cells. The stop-and-go QPI system obtains for each cell an image-based phenotype by quantifying the phase shift caused by intracellular refractive elements, such as organelles, lipid droplets, and senescent lysosomes. This system briefly stops single cells in-flow to capture long-exposure and sequential, multimodal imaging: QPI holograms for phase imaging and amplitude, as well as fluorescence labels when required. Despite stopping the cells, the system still achieves a throughput of approximately 300 to 400 cells per hour, a balance between quality and speed. The system successfully classified HEK293 cells with and without lipid droplets. A logistic regression model (LR) was trained on QPI-derived features, such as maximum phase shifts and intracellular phase variance, and the model distinguished lipid-rich cells from control cells without lipids with nearly 98% accuracy. These results show that intracellular refractive index differences captured by QPI serve as image-based markers to identify a cell state in a label-free manner. Moreover, I assessed the capacity of the stop-and-go QPI system to phenotype senescent skin fibroblasts and distinguish them from controls. Phase images of senescent cells were found to be significantly more predictive than amplitude, even after regressing out the effects of cell area, a key morphological marker of senescence. A ResNet18 neural network outperformed both LR and eXtreme Gradient Boosting (XGB), demonstrating the neural network's ability to automatically extract features from the senescence images that the engineered, feature-based models might have missed. In the future, the goal is to couple the stop-and-go microfluidic chip with 3D tomographic QPI to reconstruct volumetric refractive index maps of single cells for enhanced phenotypic analysis. Additional future directions include implementing the stop-and-go mechanism for QPI-based cell sorting, and pairing QPI and transcriptomics with single-cell resolution, broadening the scope of tools available to probe cellular heterogeneity.