Microtubules are major components of the cytoskeleton and play an important role in a number of cellular functions such as maintaining cell shape, cell division and transport of various molecules. Abnormal dynamic behavior of microtubules has been associated with neuro-degenerative diseases (e.g., Alzheimer) and cancer. Researchers study the dynamics of microtubules under different experimental conditions including different drug treatments, and using time sequence images from fluorescence microscopy. At present the dynamics of microtubules are quantified using simple first and second-order statistical measures of the length variations of manually tracked microtubules. The current analysis being mostly done manually, is quite laborious and time-consuming. Besides, the number of microtubules that one can track with manual methods is limited. In the first part of the thesis, we propose novel tools for automated detection and tracking of microtubules. A multiframe graph-based approach is proposed to tackle tracking issues, and our results demonstrate the robustness of the proposed approach to occlusions and intersections. In the second part of the thesis, we propose the use of statistical modeling tools for a better understanding of the underlying molecular mechanisms of microtubule dynamics. Prototype models are estimated for various experimental conditions by training hidden Markov models (HMMs) on the microtubule tracking data. Furthermore, these models are used to quantify similarities between experimental conditions. Additionally, temporal association rules are derived to characterize frequent patterns in the microtubule dynamics under different experimental conditions. The extraction of frequent patterns leads to a better understanding of how an experimental condition, such as the application of a drug, modulates microtubule dynamics.