3D Confocal Microscope Image Enhancement by Richardson-Lucy Deconvolution Algorithm with Total Variation Regularization: Parameters Estimation

Abstract

Deconvolution is an efficient tool for enhancing both fluorescence and confocal microscopy images. Although in confocal microscopy the point spread function is rather small and images are much sharper compared to fluorescence microscopy, deconvolution can considerably improve image contrast and reduce noise.Several deconvolution methods have been proposed for 3D microscopy. In this work, we used Richardson-Lucy (RL) iterative algorithm assuming Poisson noise. In the presence of noise, the RL algorithm does not always give a optimal solution. To reduce the effects of noise, the RL algorithm is combined with total variation regularization. As it has been shown before, total variation (TV) with a carefully chosen regularization parameter reduces intensity oscillations in homogeneous areas and helps enhance contrast on edges.The aim of this work was to find good estimates to deconvolution algorithm parameters from the input to obtain optimal results. The test images generated from experimental data by smoothing out the noise but keeping typical structures. The analysis showed that optimal algorithm parameter Lambda is strongly correlating with the noise level. On the other hand finding the optimal parameter value is a very tedious process and so we derived formula estimating the value from the input image. The estimated Lambda turned out to give a more robust stopping criterion then the popular criterion using the threshold level for the relative change between two successive iteration steps.We applied the deconvolution algorithm to study mitochondrial organization in rat cardiomyocytes. An open source software for deconvolving 3D images is available in http://sysbio.ioc.ee/software/.