BIAS FREE FEATURES DETECTION FOR HIGH CONTENT SCREENING

Recent automated confocal microscopes used in high content screening (HCS) platforms require fully automated quantitative analysis due to the large amount of images they produce. Tuning the imaging process can not be manually performed on each image, therefore these automated acquisitions may come along with strong illumination artifacts due to poor physical imaging conditions. Such artifacts obviously have direct consequences on the efficiency of the image analysis algorithms but also on the quantitative measures. In this paper, we propose a method robust to illumination artifacts to extract any kind of small and isotropic objects within cells. To do so, we include a pre-processing step where a bias correction algorithm first attempts to retrieve original images from corrupted observations. We validate our framework with two different and independent statistical criteria