Skip to main content
Fig. 2 | BMC Bioinformatics

Fig. 2

From: A cell-level quality control workflow for high-throughput image analysis

Fig. 2

Performance of cell-level QC, assay α. a Images sampled by our phenotype sampler are grouped into 7 phenotypes with 5 examples shown for each. b-c A comparison between conventional image quality measurements and ARcell. Each dot represents one image with the size and intensity proportional to ARcell. For the conventional QC metrics, FocusScore (FS) and PowerLogLogSlope (PLLS), the dashed line indicates the median value and the solid lines show 1st quartile - IQR*1.5 and 3rd quartile + IQR*1.5 respectively (IQR = interquartile range). Examples of cell-level QC result (colored arrows) are listed with their FS, PLLS and ARcell (AR) in (C). In all QC masks, artifacts are marked in green and good quality cells are labeled in blue. d Cell-level QC improves the accuracy of well-level phenotype summary. Top, an example of inconsistent dose response: out of six replicates (grey curves), the outlier curve (marked by red squares) contains images with ARcell between 3.8 and 63.0% as indicated by marker size. Bottom, comparison of drug response consistency before and after cell-level QC at concentration equal to 10uM (see Supplementary Fig. 7 for other concentrations). Each circle represents one compound

Back to article page