Robust imaging flow cytometer based on in-silico optofluidic time-stretch imaging with assistance of optical phase

Imaging flow cytometry is a widely used technique for high-throughput, label-free single-cell analysis. However, its effectiveness is often compromised by experimental perturbations, such as random defocusing and off-axis light coupling, which degrade image quality and hinder reliable analysis. In this study, we present a robust imaging flow cytometer based on in-silico optofluidic time-stretch imaging with the assistance of optical phase. We employ a generative adversarial network (GAN) trained on paired bright-field and phase images to effectively mitigate perturbation-induced artifacts. Experimental results demonstrate a ∼67 % improvement in cell classification accuracy, illustrating the enhanced robustness of the system. This approach offers a scalable and reliable method for improving the performance and accuracy of high-throughput single-cell imaging systems.