Classifying microtubule networks using curvature calculation of discrete curves

Abstract

Microtubules are structures within the cell that form a transportation network along which motor proteins tow cargo to destinations within the cell. To establish and maintain a structure capable of serving the cell's tasks, microtubules undergo deconstruction and reconstruction regularly. This change in structure is critical to tasks like wound repair and cell motility.

The method is employed on images of subpopulations of microtubules in cells to test if the subpopulations can be classified from the total population by angle distribution, curvature distribution, mean of curvature, and variance of curvature.

Images of fluorescing microtubule networks are captured in grayscale at different wavelengths, displaying different tagged proteins. The analysis of these polymeric structures involves identifying the presence of the protein and the direction of the structure in which it resides. The method presented finds statistical properties of parts of microtubules. The method processes the captured image by finding a microtubule point in the image, and determining a direction at that point. The method is then refined to estimate angular direction and curvature of the microtubules, statistically estimate the direction of microtubules in a region of the cell, and compare properties of different types of microtubule networks in the same region. To verify accuracy, we study results of the method on a test image.