Accelerating image reconstruction of asynchronous optofluidic time-stretch imaging flow cytometry

Abstract

Imaging flow cytometry is a powerful tool for profiling cell samples based on cell morphology. Optofluidic time-stretch imaging flow cytometry enhances this process with high-throughput and high-precision cell analysis capabilities, essential for fields such as cell biology and clinical medicine. The processing platform integrated with ADC and FPGA is crucial for achieving real-time acquisition and analysis of the cell image data. Synchronous sampling simplifies the image reconstruction process and provides good image quality but increases overall system complexity. Conversely, asynchronous sampling can significantly simplify the system structure and enhance flexibility, but it requires a complex algorithm to compensate for deviations between the sampling clock and the pulse repetition frequency, affecting the real-time data processing performance. In this study, we propose a calibration algorithm to accurately reconstruct the optical time-stretch imaging (OTS) image of asynchronous OTS systems. By implementing a portion of the proposed algorithm on the FPGA during the front-end data acquisition process, we achieve up to 26.82 times acceleration in image reconstruction. Furthermore, we leverage thread acceleration to enable asynchronous OTS systems to achieve a frame rate of up to 1066.11 frames per second (FPS), representing the highest reported rate for image reconstruction in asynchronous OTS imaging. This research provides valuable insights into optimizing OTS imaging systems for real-time applications, facilitating the commercialization of optofluidic time-stretch imaging flow cytometry and widespread adoption in clinical applications.