Fig. 1
From: Label-free ghost cytometry for manufacturing of cell therapy products
Schematic of a workflow for supervised machine learning in label-free ghost cytometry (LF-GC). A training dataset is prepared from the individual or combination of temporally modulated label-free “imaging” waveforms (GMI: ghost motion “imaging” signals) together with ground truth labels determining the cellular characteristics simultaneously acquired from each cell (left two columns). A machine-learning model is then trained by using the label-free GMI signals annotated with the ground truth labels in the training dataset (the third column from the left). Once the training is complete, specific cell subsets are predicted only by observing the label-free GMI signals (the rightmost column).
