Cell Classification in Human Peripheral Blood Using the Amnis ImageStream® System.

Amnis Corporation’s ImageStream® system combines the sample handling and quantitative power of flow cytometry with high-resolution brightfield, darkfield, and fluorescence cellular imagery. The system simultaneously generates up to six images of each cell in flow and can acquire data sets consisting of tens of thousands of cells in just a few minutes, while offering fluorescence sensitivity equal to or better than flow cytometry. The image data are analyzed using Amnis’ IDEAS® software, which automatically calculates over 200 morphometric and photometric features and associated statistics for each cell, identifying unique cell groups based not only on their fluorescence intensity signature but also on their morphological characteristics. The software offers the ability to view the imagery associated with any cell in a scatter plot, perform “virtual cell sorts” of user-specified sub-populations, and generate custom features of biological significance (e.g. N/C ratio). The ImageStream platform’s ability to quantitate morphologic and immunofluorescent differences between very large numbers of cells in suspension make it particularly well suited for hematology. In the present study, human peripheral blood mononuclear cells were stained with a fluorescent DNA binding dye to reveal nuclear morphology, as well as fluorescently labeled mAb to various CD markers. Five images of each cell were acquired: brightfield (transmitted light), darkfield (laser side scatter), and three fluorescent colors for nuclear imagery and quantitation of the CD marker abundance. The object of the study was to identify morphometric parameters in the brightfield, darkfield, and nuclear imagery that would prove useful in hematologic cell type classification. The mAb to CD antigens provided a positive control for use in the evaluation of the of the various morphometric parameters. Parameters with discriminating power included cellular size and texture, darkfield intensity and granularity, and nuclear fluorescence intensity, texture, and shape. Cell types that could be automatically discriminated using these parameters in lieu of immunofluorescent markers included neutrophils, eosinophils, monocytes, and lymphocytes (including putative activated lymphocytes). In addition to forming the basis for an advanced ImageStream hematology platform, it is envisioned that the automated morphometric classification of blood cells will act as the foundation for a wide range of image-based cellular assays performed in peripheral blood (e.g. NF-kB translocation, apoptosis, mAb compartmentalization), allowing the differential quantitation of assay results in various cell types for the purposes of basic research, drug discovery, and clinical diagnostics.