Deconvolving Active Contours for Fluorescence Microscopy Images

We extend active contours to constrained iterative deconvolution by replacing the external energy function with a model-based likelihood. This enables sub-pixel estimation of the outlines of diffraction-limited objects. such as intracellular structures, from fluorescence micrographs. We present an efficient algorithm for solving the resulting optimization problem and robustly estimate object outlines. We benchmark the algorithm on artificial images mid assess its practical utility on fluorescence micrographs of the Golgi and endosomes in live cells.