Whole Organism High-Content Screening by Label-Free, Image-Based Bayesian Classification for Parasitic Diseases
Original images were firstly segmented as described Figure 1 and then the cumulative change in area for each individual larva measured over 5 time frames. The scores for individual larvae were averaged to generate a motility score per well. (A) Illustration of differences in larval motility between a high motility well (DMSO) with low overlap between larval location in successive time frames (red boundaries), a medium motility well (PZQ; praziquantel) and low motility wells with almost perfect overlap between successive time frame boundaries (DHA; dihydroartemisinin & OPZ; oltipraz). (B) Representative motility scores from 5 plates of anti-schistosome compounds (10 µg/ml), n = 120 wells per condition and significance values were measured by non-parametric one-way ANOVA with individual drug treatments being compared to the DMSO controls wells using a Dunn's comparison post test, *** = p<0.001. NS; not significant. (C) Z factor analysis based on the distribution of the negative (DMSO; open circle) and positive controls (OPZ; closed triangle) from 15 plates (dotted lines represent upper and lower thresholds).