CIVIT dataset: Integral microscopy with Fourier plane recording

Area of Science:

• Microscopy and Imaging
• Computational Biology
• Computer Vision

Background:

• Fourier Lightfield Microscopy (FLMic) offers advanced 3D imaging capabilities.
• Development of robust volumetric reconstruction algorithms is crucial for analyzing FLMic data.
• Synthetic datasets are valuable for algorithm development and validation in microscopy.

Purpose of the Study:

• To present a comprehensive synthetic dataset for Fourier Lightfield Microscope (FLMic) imaging.
• To facilitate the development and assessment of volumetric reconstruction techniques for FLMic data.
• To provide researchers with standardized data for benchmarking light field microscopy algorithms.

Main Methods:

• Generation of 22,416 synthetic images using Blender, simulating biological samples (cells, bacteria).
• Creation of eight distinct scenes with 3D models, including 'Cells' and 'Filaments' categories.
• Provision of two data structures: 51x51 Elemental Images (EIs) for Densely Sampled Light Fields (DSLF) and Z-Scans.

Main Results:

• A diverse dataset comprising 22,416 synthetic images and associated metadata.
• Inclusion of camera calibration parameters and MATLAB scripts for pose compensation.
• Dataset covers eight scenes with varying biological sample representations.

Conclusions:

• The presented dataset is a valuable resource for advancing volumetric reconstruction from FLMic images.
• It enables rigorous testing and comparison of novel light field imaging algorithms.
• Facilitates research in biological imaging and computational microscopy.