Handheld and Cost-Effective Fourier Lightfield Microscope

CIVIT dataset: Integral microscopy with Fourier plane recording

This article describes a dataset of synthetic images representing biological scenery as captured by a Fourier Lightfield Microscope (FLMic). It includes 22,416 images related to eight scenes composed of 3D models of objects typical for biological samples, such as red blood cells and bacteria, and categorized into Cells and Filaments groups. For each scene, two types of image data structures are provided: 51 × 51 Elemental Images (EIs) representing Densely Sampled Light Fields (DSLF) and 201 images composing Z-Scans of the scenes. Auxiliary data also includes information about camera intrinsic and extrinsic calibration parameters, object descriptions, and MATLAB scripts for camera pose compensation. The images have been generated using Blender. The dataset can be used to develop and assess methods for volumetric reconstruction from Light Field (LF) images captured by a FLMic.

Resolution limit in opto-digital systems revisited

The resolution limit achievable with an optical system is a fundamental piece of information when characterizing its performance, mainly in case of microscopy imaging. Usually this information is given in the form of a distance, often expressed in microns, or in the form of a cutoff spatial frequency, often expressed in line pairs per mm. In modern imaging systems, where the final image is collected by pixelated digital cameras, the resolution limit is determined by the performance of both, the optical systems and the digital sensor. Usually, one of these factors is considered to be prevalent over the other for estimating the spatial resolution, leading to the global performance of the imaging system ruled by either the classical Abbe resolution limit, based on physical diffraction, or by the Nyquist resolution limit, based on the digital sensor features. This estimation fails significantly to predict the global performance of opto-digital imaging systems, like 3D microscopes, where none of the factors is negligible. In that case, which indeed is the most common, neither the Abbe formula nor the Nyquist formula provide by themselves a reliable prediction for the resolution limit. This is a serious drawback since systems designers often use those formulae as design input parameters. Aiming to overcome this lack, a simple mathematical expression obtained by finely articulating the Abbe and Nyquist formulas, to easily predict the spatial resolution limit of opto-digital imaging systems, is proposed here. The derived expression is tested experimentally, and shows to be valid in a broad range of opto-digital combinations.

Machine Learning-Based View Synthesis in Fourier Lightfield Microscopy

Current interest in Fourier lightfield microscopy is increasing, due to its ability to acquire 3D images of thick dynamic samples. This technique is based on simultaneously capturing, in a single shot, and with a monocular setup, a number of orthographic perspective views of 3D microscopic samples. An essential feature of Fourier lightfield microscopy is that the number of acquired views is low, due to the trade-off relationship existing between the number of views and their corresponding lateral resolution. Therefore, it is important to have a tool for the generation of a high number of synthesized view images, without compromising their lateral resolution. In this context we investigate here the use of a neural radiance field view synthesis method, originally developed for its use with macroscopic scenes acquired with a moving (or an array of static) digital camera(s), for its application to the images acquired with a Fourier lightfield microscope. The results obtained and presented in this paper are analyzed in terms of lateral resolution and of continuous and realistic parallax. We show that, in terms of these requirements, the proposed technique works efficiently in the case of the epi-illumination microscopy mode.