TRACING MICROTUBULES IN LIVE CELL IMAGES

Microtubule (MT) dynamics are traditionally analyzed from time lapse images by manual techniques that are laborious, approximate and often limited. Recently, computer vision techniques have been applied to the problem of automated tracking of MTs in live cell images. Aside of very low signal to noise ratios, live cell images of MTs exhibit severe clutter for accurate tracing of MT body. Moreover, intersecting and overlapping MT regions appear brighter due to additive fluorescence. In this paper, we present a MT body tracing algorithm that addresses the clutter without imposing directional constraints. We show that MT dynamics can be quantified with enhanced precision, and novel measurements that are beyond manual feasibility, can be obtained accurately. We demonstrate our results on actual images of MTs obtained by live cell fluorescence microscopy.