A framework for generating realistic synthetic sequences of dynamic confocal microscopy images

In recent years many automated methods for detection and tracking of sub cellular structures in live cell fluorescence microscopy have been proposed. Because dependable ground truth from real data sets is difficult to obtain, most algorithms are tested on synthetic data where the ground truth is known. Differences between real and synthetic data sets can lead to imprecise judgement about an algorithm’s performance. In this paper we present a method for generating realistic synthetic sequences of live cell confocal microscopy images that simulate the image formation as well as modelling the motion
of dynamic structures during image acquisition using valid dynamic models. Sequences generated using this framework realistically reproduces the complexities existing in real confocal microscopy sequences.