An integrated approach to determine prior information for improved wide-field imaging models from computational interference microscopy

摘要

Knowledge of prior sample information, such as a refractive index (RI) map, can be used to improve image formation models enabling more accurate three-dimensional (3D) restoration in fluorescence microscopy. RI is an indicator of cell composition and structure that allows a more comprehensive representation of the 3D structure of a specimen than fluorescence alone. Due to the integral nature of sample phase, the challenge to compute the RI map is to decouple RI and thickness. Our work investigates the feasibility of determining RI of a specimen from differential interference contrast (DIC) microscopy data acquired by using different wavelengths in illumination. This spectral diversity in the data is exploited to determine sample thickness and RI. Results from simulated and experimental data of polystyrene bead samples are presented to analyze this approach. Phase images were estimated from the DIC data using an alternating minimization algorithm. This study shows that the maximum estimated phase delay is accurate within approximately 7 percent error relative to the 2D phase model. The sensitivity of this integrated approach allows RI to be computed within approximately 0.4 percent error relative to values from the literature.