For decades, flow cytometry has been recognized as the classic technique for single-cell analysis. Used to detect and characterize distinct—and often incredibly rare—cell types within large, heterogeneous populations, it has seen widespread use to support a diverse range of research applications. However, despite being an extremely fast, sensitive, and quantitative approach to cellular identification, a major limitation of flow cytometry is that it provides no morphological or spatial resolution. This restricts its utility to intensity-based analysis, leaving researchers in the dark when it comes to establishing exactly where in a cell a particular signal originates from.
To gain deeper insights from sample that is often available in only short supply, many researchers are turning to imaging flow cytometry (IFC) as a preferred alternative to standard flow cytometry techniques. By combining the high-throughput, multiparametric analysis capabilities and statistical significance of flow cytometry with the morphological and spatial resolution of microscopy, imaging flow cytometry provides researchers with the ability to capture multiple digital images of many thousands of individual cells in just minutes. This information can be used to create highly detailed cellular profiles, allowing comparison with other cell types to interrogate specific sub-populations or to establish links between cellular phenotype and disease.
Recommended eBook
Download this ebook now to learn about new solutions that make IFC more accessible than ever.One of the current challenges of analyzing the vast data sets generated by imaging flow cytometry is the level of expertise necessary to perform complex masking and feature calculation. In this ebook, we explain how leveraging the power of machine learning (ML) and artificial intelligence (AI) can help researchers more effectively analyze their imaging flow cytometry data, before delving into some specific use cases. These include using AI to support micronuclei detection and to measure the immunological synapse and applying machine learning to quantify white blood cells (WBCs).
Benefits of imaging flow cytometry
Unlike conventional flow cytometry, which typically uses fluorescence intensity as a measure of distinct cell surface markers, imaging flow cytometry combines brightfield, darkfield, and fluorescence-based detection all in one platform. This is achieved using 20×, 40×, or 60× objectives in addition to a unique time delay integration (TDI) charge-coupled device (CCD) camera. Although both techniques operate similarly, imaging flow cytometry not only acquires fluorescence intensity but also provides detailed imagery of every cell within a sample. To view morphological and structural cellular properties, researchers simply select any dot within a dot plot to look more closely at an individual cell or select a specific bin within a histogram to see all the cells within a defined sub-population.
Depending on the number of fluorescent channels the imaging flow cytometer has available, it is possible to capture as many as 12 digital images of each cell (brightfield and darkfield images, plus 10 fluorescent readouts), translating to hundreds of thousands of images per sample. The depth of information on a cell-by-cell basis is comparable to that of standard microscopy. However, because imaging flow cytometry also benefits from the statistical significance of large sample sizes common to conventional flow cytometry, it is considerably more powerful than either technique used alone. These features of imaging flow cytometry make it well-suited to a broad range of applications, including studies designed to monitor multiple subcellular compartments, or to locate and quantify the distribution of signaling molecules on, in, or between cells.
Challenges of imaging flow cytometry
Although imaging flow cytometry provides extraordinarily rich morphological and spatial information, handling such vast quantities of data presents significant challenges. For want of a better approach, analyses are often based on just a small number of selected features, many of which are identified manually by applying binary gates. While such a strategy can work in the hands of a researcher experienced in imaging flow cytometry analysis, it is highly prone to user bias and requires considerable interaction with the data. Moreover, by failing to consider all the available information, manual analysis of imaging flow cytometry data can lead to valuable insights being overlooked.
New frontiers in analyzing IFC data
Recently, artificial intelligence, and specifically machine learning, have driven huge advances within the field of bioimaging research. Machine learning uses algorithms to identify and separate distinct cellular populations. An example is the linear discriminant analysis (LDA) that creates unique classifiers based on optimally combining multiple image characteristics and morphologies into population-specific features. AI, which utilizes deep learning algorithms, exploits the opportunities afforded by user guided training and computer vision to further simplify multimodal image analysis, thereby reducing user-related variability and improving the overall quality and reliability of imaging flow cytometry data analysis.
As well as helping to reveal findings that would otherwise remain hidden within an overwhelming wealth of experimental readouts, machine learning and AI have been fundamental in increasing the uptake of imaging flow cytometry. In turn, this has paved the way to the development of novel experimental capabilities, consequently opening up a greater diversity of applications. Areas where imaging flow cytometry has proven especially valuable include phenotyping and identifying circulating tumor cells, studying cell-cell interactions, and monitoring disruptions to cell signaling mechanisms.
To make imaging flow cytometry more accessible, Luminex offers two imaging flow cytometers (the Amnis® ImageStream®X Mk II and the Amnis® FlowSight®) as well as two different software options (IDEAS® 6.3 plus Machine Learning and the Amnis® AI Image Analysis Software, both of which are compatible with either instrument). We will now consider some specific cases where the Amnis® platforms have been deployed: the adaptation of a cytokinesis-block micronucleus (CBMN) assay to an imaging flow cytometry format for improved throughput; measurement of the immunological synapse (a rare entity that has historically been difficult to analyze objectively), using FlowSight® imagery for immunological synapse identification; and development of a method for faster, more accurate quantification of WBCs.
To learn more about the benefits of IFC as well as new ML- and AI-powered solutions to the current data analysis challenges posed by its vast data sets, download our free eBook now.