DIC Microscopy Image Reconstruction Using a Novel Variational Framework

Abstract

Quantitative microscopy (QM) became a key tool in systems-level drug discovery and disease diagnosis such as cancers and neurodegenerative disorders. However, to date QM is limited to epifluorescence microscopy which requires chemical labels, special imaging modality and often causes phototoxicity. Differential Interference Contrast (DIC) microscopy is label free and is low-phototoxic, thus it has great advantages over epifluorescence microscopy in numerous applications. Yet, DIC is not used for QM because the acquired images are not feasible directly for quantitative analysis. In this paper we propose a novel variational framework for DIC image reconstruction, enabling the modality for QM. Our energy functional uses a term that ensures similarity to the original DIC image and the total variation regularization term. The first term utilizes the point spread function (PSF) of the DIC microscope. The PSF is incorporated to our model by local integrals. We show that the derivation operation can be moved from the kernel to the image, which significantly accelerates the computations. The method outperforms other algorithms on synthetic and real DIC images.