Fig. 2: Lipid droplet dataset.
From: Regression plane concept for analysing continuous cellular processes with machine learning
a Training set. Regression plane of 457 cells representing various lipid morphologies, created by an expert biologist. b RP output. Kernel Density Estimation (KDE)-maps of the predicted regression positions for cells treated with selected siRNAs. Arrows originate from the peak of the control KDE-map, and point to the peaks of the selected KDE-maps. c HCS analysis. Plate-based analysis performed by comparing well-based KDE-maps. Meta-visualization (in this case PCA–Principal Component Analysis) is obtained by extracting the principal components (PC1 and PC2) of the flattened KDE-maps.
