Investigation of the microtubule dynamics with probabilistic data association filter

Abstract

Understanding microtubule dynamics has important implications for establishing nanometer level machines. Object tracking is one of the important issues necessary to elucidate the dynamics of microtubule from video data. In microtubule gliding assays, object tracking becomes non-trivial due to the occurrences of compound objects and high density objects. In this work, we investigate microtubule dynamics focusing on its morphological information, and we developed easy and useful workflow with compound segmentation technique and probabilistic data association filter. Using this workflow, multi-crossing microtubules can be decomposed, and be tracked correctly.