Figure 9.

Categorization of microglial cells by means of a logical decision tree based on morphological parameters. (A) With the aim of assigning cells into one of the four proposed clusters a logical decision tree was designed based on the strongest predictor parameters revealed by the linear discriminant analysis (LDA). For each individual microglial cell, the first parameter to be analyzed would be the CHSR. Cells with a value greater than 1.93 could be assigned to Cluster 4, while those with lower values would be then evaluated by its CC or its CHA. These two parameters would allow to allocate cells to Clusters 1, 2 and 3, according to the parameter values indicated in (A). (B) Principal components analysis (PCA) suggested a further microglia classification with a descriptive categorization. In this case cells within the different clusters were classified based on values respect to the plot axis in the PC plane. The main parameter of PC-1, MSACH, allowed to split Clusters 1 and 3, resulting in Types 1.1 and 1.3, and Types 3.1 and 3.2, respectively. In both cases, the resulting Types were significantly different according to their MSACH values (P < 0.001). The main parameter of PC-2, CHC, was used to split Clusters 2 and 4 in Types 2.1 and 2.2 and Types 4.1 and 4.2, respectively. Differences in CHC between these Types were significant (P < 0.001). Cells are all in the same scale (scale bar = 50 μm).