Imaging flow cytometry (IFC) marries the images of microscopy with the throughput of flow cytometry. But pushing throughput up forces trade-offs: richer imaging (fluorescence, phase) usually means slower — often giving up continuous streaming or real-time analysis — while faster usually means giving up fluorescence. This series maps the main high-throughput IFC techniques — and, importantly, which real instrument embodies each — based on the comparison table in Zhou et al., Light: Science & Applications (2025), updated with the current commercial landscape. This first post covers the two foundational approaches: conventional IFC and optical time-stretch (OTS).
成像流式细胞术（IFC）把显微镜的成像能力与流式细胞术的高通量结合在一起。但要提高通量，就必须权衡取舍：更丰富的成像（荧光、相位）通常意味着更慢，往往要牺牲连续数据流或实时分析；而更快通常意味着放弃荧光。本系列梳理主要的高通量 IFC 技术，并重点说明每种技术对应的真实仪器 —— 基于 Zhou 等人发表于 Light: Science & Applications (2025) 的对比表，并补充当前的商业化情况。本篇（第一部分）先介绍两种基础方法：传统 IFC 与光学时域拉伸（OTS）。
The six techniques / 六种技术
1. Conventional IFC / 传统成像流 式
Conventional IFC uses standard lasers or LEDs with high-resolution CCD/CMOS sensors to deliver multi-color fluorescence,a spatial resolution of 0.3-1um and quantitative phase imaging at roughly 1,000 eps. Its real-time image analysis is the most mature of any technique here, but its speed is physically capped by camera exposure and readout.
传统 IFC 采用常规激光或 LED 光源，结合高分辨率的 CCD/CMOS 相机（图像传感器），在每秒约一千个细胞下提供多色荧光与定量相位成像，0.3-1um的空间分辨率。它的实时图像分析实现度是所有技术中最成熟的，但其速度受限于相机曝光与读出的物理极 限。
Instrument / 代表仪器: Cytek Amnis ImageStreamX Mk II / FlowSight (CCD in TDI mode; originally Amnis → Luminex → now Cytek Biosciences).
2. OTS — optical time-stretch / 光学时域拉伸
OTS (optical time-stretch imaging) is the technology platform advanced in this paper by Cheng Lei’s group at Wuhan University. Rather than relying on a conventional camera, OTS converts spatial information into a serial time-domain signal that can be read out by a single high-speed photodetector.
A broadband ultrashort laser pulse is first dispersed into a “rainbow,” in which each wavelength corresponds to a different position along a one-dimensional illumination line. As a cell flows through this rainbow line in a microfluidic channel, its morphology modulates the intensity of each wavelength at once, encoding the cell’s spatial image into the optical spectrum.
The encoded rainbow light is then sent through a long dispersive optical fiber. Because different wavelengths travel at slightly different speeds in the dispersive medium, they no longer arrive together but emerge one after another in time. The spatial information that originally existed all at once is thus transformed — after detection by a single-pixel photodetector — into a time-ordered electrical waveform.
Each laser pulse captures one line scan of the moving cell; as the cell continues flowing, successive line scans are stacked in chronological order to reconstruct a two-dimensional image.
Unlike conventional imaging flow cytometers, OTS does away with the concept of camera exposure entirely. Image acquisition instead uses an 80 MHz broadband pulsed laser, a 12 GHz single-pixel photodetector, and a 10 GS/s ADC, with real-time image reconstruction and processing performed on an FPGA. Under the conditions reported in this work, the system achieved a real-time throughput exceeding 1,000,000 events per second (eps), supported cell velocities up to 15 m/s, and delivered a spatial resolution of approximately 780 nm at the 10 GS/s sampling rate.
Because OTS relies on wavelength-to-time mapping, it is inherently suited to bright-field / label-free imaging and is generally not compatible with fluorescence.
OTS natively integrates quantitative phase imaging (QPI), allowing it to calculate biophysical traits like cellular dry mass and thickness completely label-free.
A major engineering challenge of OTS is the sheer data rate it generates. In this system, recording every pulse produced a front-end data stream approaching 4.8 GB/s, while the raw ADC output reached roughly 20 GB/s. To bring the bandwidth down to a level the downstream data bus and storage can sustain, the FPGA performs real-time background subtraction, region-of-interest (ROI) detection, 16-bit-to-8-bit requantization, and redundant-pulse skipping before the data is transferred.
Beyond the system in this paper, another representative OTS research platform comes from Professor Keisuke Goda’s group at the University of Tokyo — one of the pioneers of optical time-stretch imaging flow cytometry.
OTS（光学时域拉伸成像）是本篇论文（武汉大学雷诚团队）所推进的前沿技术流派。它不依赖传统相机，而是把空间信息转换成一串随时间排列的信号，由单个高速光电探测 器读出。
打个比方：OTS 先把一束宽带超短激光脉冲“散成彩虹”，让每一种波长对应照明线上的一个位置（如 下示意）：
450nm 650nm
────
────
────
────
│ │ │ │ │
左侧 左中 中间 右中 右侧
当细胞在微流控通道里流过这条“彩虹线”时，每一种波长只照亮细胞上的一个位置，因此“空间位置”被编码成了 “ 波长”。
带着细胞信息的彩虹光，再排队穿过一根很长的色散光纤。由于不同颜色在色散介质里走得快慢不同，它们不再挤在同一瞬间，而是先后到达探测器。这样本来“同时存在的空间信息”，经单像素探测器接收后，就变成了“按时间排队的一 串电信号”。
每个激光脉冲捕获细胞的一次线扫描（一个截面）；随着细胞继续流动，这些线扫描按时间顺序依次堆叠，就重建出 一幅二维细胞图像。
它彻底颠覆了传统相机的曝光概念：系统使用 80 MHz 宽带脉冲激光、12 GHz 单像素探测器和 10 GS/s ADC，并由 FPGA 完成实时图像重建与处理。在本文的实验条件下，系统实现了超过 1,000,000 eps 的实时检测通量，细胞流速最高达 15 m/s，并在 10 GS/s 采样率下实现了约 780 n m 的空间分辨率。
由于 OTS 依赖“波长—时间”映射，它天生适合明场 / 无标记 成像，不兼容荧光成像。
QTS成像技术自带定量相位成像(QPI)功能，在无标记情况下直接计算出细胞的厚度和干质量等重要的生物物理特征。
OTS 的一大工程挑战是它产生的巨大数据率。在该系统中，若记录每一个脉冲，前端数据流接近 4.8 GB/s，而 ADC 原始输出更高达约 20 GB/s。为把带宽压缩到下游数据总线与存储系统可承受的范围，FPGA 会先进行实时背景去除、感兴趣区（ROI）检测、16 bit 到 8 bit 重新量化，以及冗余脉 冲跳过，然后再传输数据。
除本文外，该技术流派的另一代表性科研平台来自日本东京大学 Keisuke Goda 教授团队——光学时间拉伸 成像流式细胞术的先驱之一。
Instrument / 代表系统: Lab-built — OTS has not yet been incorporated into a commercial imaging flow cytometer. / 实验室自建 — 目前尚无商业化机型。
Based on Zhou et al., “Imaging flow cytometry with a real-time throughput beyond 1,000,000 events per second,” Light: Science & Applications 14:76 (2025),