Simple computational technique to quantify nuclear shape asymmetry.

Abstract

The nucleus of an eukaryotic cell is a membrane-bound organelle containing a major part of the cellular genome. Nuclear shape is controlled by forces generated in the cytoskeleton, nuclear envelope and matrix of the nucleus and may change when the balance of these forces is disturbed. In certain cases, such changes may be indicative of cell pathology. Nuclear shape feature is being commonly addressed in both experimental research and diagnostics; nevertheless its symmetry-related aspects receive little attention. This article introduces a technique allowing to estimate nuclear shape asymmetry in digital images captured from cyto- or histological preparations. Implemented in a software package, this technique quantifies the asymmetry using two scenarios. The first one presumes the identification of nuclear pixels laying outside the largest inscribed circle. According to the second scenario, the algorithm searches for nuclear pixels lacking pixel-partners symmetric with respect to the nuclear area's centroid. In both cases, the proportion of "asymmetric" pixels is used to estimate the feature of interest. The technique was validated on images of cell nuclei having distinctive shape phenotypes. A conclusion was made that shape asymmetry feature may be useful accessory to the toolbox of nuclear morphometry.