3D Anisotropic Diffusion Filtering for Enhancing Noisy Actin Filament Fluorescence Images

Abstract

We present a PDE-based method for increasing the S/N ratio in noisy fluorescence image sequences where particle motion has to be measured quantitatively. The method is based on a novel accurate discretization of 3-D non linear anisotropic diffusion filtering, where the third dimension is the time t in the image sequence, using well adapted diffusivities. We have applied this approach to fluorescence image sequences of in vitro motility assay experiments, where fluorescently labelled actin filaments move over a surface of immobilized myosin. The S/N ratio can be drastically improved resulting in closed object structures, which even allows segmentation of individual filaments in single images. In general this approach will be very valuable when quantitatively measuring motion in low light level fluorescence image sequences used in biomedical and biotechnological applications for studying cellular and subcellular processes and in in vitro single molecule assays.