{"id":3442,"date":"2025-03-10T21:42:05","date_gmt":"2025-03-10T20:42:05","guid":{"rendered":"https:\/\/websites.fraunhofer.de\/smart-sensing-insights\/?p=3442"},"modified":"2025-10-14T11:06:18","modified_gmt":"2025-10-14T09:06:18","slug":"mikaia-cellular-neighborhood-app","status":"publish","type":"post","link":"https:\/\/websites.fraunhofer.de\/smart-sensing-insights\/mikaia-cellular-neighborhood-app\/","title":{"rendered":"MIKAIA Cellular Neighborhood App Explained"},"content":{"rendered":"\n

The MIKAIA\u00ae<\/sup> Cellular Neighborhood App can yield statistics about the cellular microenvironment<\/strong>, i.e., identify cell niches<\/strong> or cellular communities<\/strong>. This presents a type of spatial analysis of cell populations detected by the FL Cell Analysis App<\/a> or H&E Cell AI App, as well as those imported from annotation files (e.g., GeoJson). Additionally, the app can be integrated into comprehensive workflows, such as analyzing a 47-plex spatial proteomics scan generated with MACSima\u2122<\/a> (by Miltenyi Biotec).<\/p>\n\n\n\n

The underlying algorithm centers in on each cell and then collects what cell phenotypes are present in the current center cell’s neighborhood. The neighborhood can either be defined by a fixed radius (e.g., 50 \u00b5m), it can be comprised of the k-nearest neighbors, or a combination of both.<\/p>\n\n\n\n

The Cellular Neighborhood App then can do two things:<\/p>\n\n\n\n

    \n
  1. Yield statistics about the average neighborhood composition by cell type and ROI.<\/li>\n\n\n\n
  2. Re-classify each cell based on its cellular microenvironment<\/strong>, i.e., identify cell niches<\/strong> or cellular communities<\/strong>.<\/li>\n<\/ol>\n\n\n\n

    In this app note, these analyses and the resulting diagrams are explained in detail, based on two artificial examples.<\/p>\n\n\n\n

    Example 1<\/h2>\n\n\n\n

    The following screenshot shows a field of view that comprises three cell types (“class A”\/yellow, “class B”\/blue, “class C”\/orange).
    The cells are arranged in four zones, from top to bottom: <\/p>\n\n\n\n

      \n
    1. Yellow & blue mixed (top)<\/li>\n\n\n\n
    2. Blue only (2nd<\/sup> from top)<\/li>\n\n\n\n
    3. Orange & blue mixed (3rd<\/sup> from top)<\/li>\n\n\n\n
    4. Orange cells only (bottom)<\/li>\n<\/ol>\n\n\n\n
      \"field<\/figure>\n\n\n\n

      Example 1 — Neighborhood composition<\/h3>\n\n\n\n

      The following diagram shows the result of this analysis, when defining a neighborhood of max. 5 cells within at most 30 \u00b5m radius. It shows for each cell type (A, B, and C) a distance histogram (configured to 3 bins \u00e1 10 \u00b5m) that states the relative neighborhood cell composition in 3 bins of 10 \u00b5m each. Since the absolute area of each distance bin increases, the composition is displayed as a normalized stacked bar chart. It states both the absolute and relative average cell abundance.<\/p>\n\n\n\n

      \"stacked<\/figure>\n\n\n\n

      Example 1 — Identification of cellular neighborhoods<\/h3>\n\n\n\n

      Since “Cluster” with k=4 was enabled, 4 new cellular neighborhood annotation classes “CN-1” — “CN-4” were created and each cell was assigned to one of these classes (the original cell classes remain unchanged, but are hidden). The clustering algorithm (k-means) identified the four different zones, described above.<\/p>\n\n\n\n

      \"Four<\/figure>\n\n\n\n

      Additional diagrams summarize findings about these cellular neighborhoods. The “Abundance by ROI” simply states the abundance per type and ROI. In this simple example, the analyzed region was not divided into multiple regions of interests. <\/p>\n\n\n\n

      \"<\/figure>\n\n\n\n

      The two “Cluster Explorer” diagrams give insights on how the individual cellular neighborhoods (CN) are composed. The following diagram shows a 2D heatmap, where each row relates to a CN and each column to one of the original cell types.<\/p>\n\n\n\n

      \n
      \n
      \""Cluster<\/figure>\n<\/div>\n<\/div>\n\n\n\n

      From it, we learn how the 4 CNs are composed on average. The average cell abundances per row sum up to 6, since they consider the center cell + the 5 neighbors:<\/p>\n\n\n\n