Semisupervised Clustering by Iterative Partition and Regression with Neuroscience Applications

Figure 5

(a) Raw spatial data on 128 colorectal cancer cells imaged on a high-content microscope imager (Operetta, Perkin Elmer). (b) Spatial distribution of the 128 cells represented by the colored circles. The circles have 5 colors representing 5 clusters which resulted from LS multivariate regression clustering minimizing (14). The size of each circle estimates the area of the corresponding cell. The clustering uses RGB intensities as the response vector and cell area as the predictor. (c) 3D-scatterplot of the clustered RGB intensities of the 128 cells. Colors of the points show the same 5 clusters shown in (b). (d) Spatial distribution of the 128 cells, with the colored circles showing the 5 clusters given by the LS multivariate regression clustering not including any predictor.

(a)

(b)

(c)

(d)