FIGURE 1.

Schematic overview of the proposed method. The raw fcs files should first be preprocessed according to the suggested workflow in the blue box, which can include the removal of margins, compensation, and transformation. The method itself starts with a preprocessed fcs file. First, the sample is binned and density peaks are determined, clustered, and filtered per bin. The cleaning part of the method starts with the usage of an isolation tree, then it will remove peaks based on their MAD distance and it ends by connecting disjointed removed regions. The method returns a cleaned fcs file [Color figure can be viewed at wileyonlinelibrary.com]