81 3 Towards Automatization of Organoid Analysis: A Deep Learning Approach to Localize and Quantify Organoids Images Fig. 2. Evaluation results. (A) Representative bright field image of a liver progenitor organoid culture at day 3 after passaging with medium to small structures. 95 organoids have been selected and manually measured (indicated by red contour). Scale bar, 1000 μm (B) Size of 95 individual organoids measured manually compared to the same organoids measured by OrganelX. (C) Representative bright-field images of liver progenitor organoids imaged at time point 0 H and 96 H. The upper row shows control organoids in complete condition medium and the bottom row shows organoids in a medium depleted of amino acids (Starvation). The scale bars present 1000 μm (D) Cross-sectional area of 25 random organoids tracked in time, measured every 24 H. The black and solid line represents control conditions measured manually. The grey and dotted line represents control conditions measured by OrganelX. The red and solid line represents starvation conditions measured manually. The light red and dotted line represents starvation conditions measured by OrganelX. Error bars indicate the SEM. (*P< 0.05, **P< 0.01, *** P< 0.001, unpaired t test at individual time points for comparison between Control and Starvation conditions of the manually selected organoids. (#P< 0.05, ##P< 0.01, ### P< 0.001, unpaired t test at individual time points for comparison between Control and Starvation conditions of organoids measured by OrganelX). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) Segmentation and growth area over time Once the ability of the system to replicate manually annotated data was assessed, its ability to track organoids in time was evaluated. In this case study, liver progenitor organoids have been grown in either complete medium (referred to as control) or in medium lacking all amino
RkJQdWJsaXNoZXIy MTk4NDMw