Figure 5
From: Label-free prediction of cell painting from brightfield images
A systematic breakdown of correlation between morphological profiles. The features selected from the CellProfiler pipeline are presented as correlation heatmaps in (a) (U-Net) and (b) (cWGAN-GP). The correlations for each model are with the features extracted from the ground truth Cell Painting images and are aggregated by channel and feature group for cell, cytoplasm, and nuclei objects. The number of features for each feature group is also presented. U-Net mean = 0.43, cWGAN-GP mean = 0.45.
