Figure 1.

Schematic view of the workflow in CellClassifier. As an input, the tool needs the original microscopy images, as well as output files and label matrix images produced by CellProfiler. The interaction with the user happens mainly via the main window (for a screenshot see Figure 3). For special tasks like adaptation of the focus exclusion routine, Support Vector Machine model generation and alternative training as well as adjustment of the output and settings (not shown) more graphical user interfaces open. CellClassifier produces several different outputs: a) Outlined images with projections of the object borders, colors of the outlines thereby projecting classification or other object information (for examples see Figures 2 and 4). b) Excel-files with user defined information about the image, wells and the whole plate. c) Heat maps or other graphical summaries containing user defined overviews over a whole 96- or 384-well plate. d) Manual cell count, generating a well based summary of the classifications by the user and e) Matlab summaries, making export as Matlab variables possible.