Quantitative imaging of cell dynamics Parallelized contact imaging and automated analysis of cell migration dynamics
Résumé
We show the capacity to carry out screens for cell migrations using 96 wound healing assays achieved in 96-well microtiter plates based on a new optical technique developed by our group. Dynamics of the wound closures were obtained by parallelized time lapse contact imaging microscopy and dedicated automated image analysis. Our plate reader relies on an array of 96 image sensors, namely the planar arrangement of 12×8 image sensors, placed under the transparent flat-bottomed 96-well microtiter plate so that each well can be imaged by the image sensor placed underneath. Week-long monitoring of live cell populations showed long-term imaging position stability and no focus drift in any image series, which makes our time-lapse plate reader very competitive in comparison to conventional video microscopy equipment. The 96 wound closure dynamics were extracted from the images using a specifically developed automated segmentation method. Robust localization of the wound edges in low contrast images was achieved by global segmentation algorithms based on Markov random fields and active contours even with non-uniform illumination conditions. A parallel double snake was used to model the approximate parallelism between the two edges of a wound. The performance of global segmentation was validated on a set of images showing wounds in confluent epithelial cell cultures. Automated wound localization was compared with manual segmentation performed by seven cell biology experts by determining the root-mean-square error between the segmented interfaces and region-oriented analysis. Evaluations of intra and inter-biologist variabilities showed that automated segmentations are as accurate and robust as the cell biologist's ones. Wound closure dynamics was applied to measure and compare the motility of four affiliated prostate cell lines representing various grades of prostate cancer development.