This paper presents a computer vision framework for detecting and tracking diffraction images of linear structures in differential interference contrast (DIC) microscopy. The tracker can resolve image displacements of 1/10 of a pixel despite the weak and orientation-dependent contrast in DIC, as well as the variable blur in such image data caused by vertical specimen movement. In our high numerical aperture, high magnification microscope set-up, this resolution corresponds to 5 nm in object space. In video DIC similar resolution has been reported hitherto only for rotationally symmetric targets such as bead images. The tracker was developed for measuring deflections of clamped microtubules with a freely moving second end. By analysing the thermal fluctuations of such microtubules it was possible to derive their elasticity. The paper describes a filtering scheme for the detection and localization of DIC diffraction line images which represent loci of microtubules. For tracking the movements of the extracted lines we adopted the sum of squared (brightness) differences algorithm from computer vision. The analysis of the fluctuation measurements demonstrates the high sensitivity of this tracking technique in quantifying positional and orientational changes. We derived that the theoretical limit in tracking displacements of such diffraction line images is 1.25 nm, four times below the experimentally verified sensitivity. This indicates that the proposed tracker is still suboptimal. Nevertheless, the tracking precision was sufficient to reveal subtle deviations in the distribution of microtubule deflection from free diffusion. They were induced by pivotal points and multiple positions of relaxation. Also, the results suggest that there were defects in the polymer structure which caused very small but significant bends in the microtubule axis.