Raw and processed VIPA-based Brillouin microscopy data for physics-informed zero-shot denoising

This dataset corresponds to the paper "Physics-informed zero-shot denoising for VIPA-based Brillouin microscopy", including raw spectral data collected by VIPA spontaneous Brillouin microscope, frequency calibration and noise calibration data, ZEN denoising results, Brillouin shift diagram and quantitative evaluation data. The original data were collected by sCMOS camera, and each spectral channel in biological samples was composed of 20 non-dispersive pixel rows, which were divided into two data subsets with 10 rows each according to odd and even rows, which were used to construct zero-sample self-supervised training pairs.

Noise is described by Poisson-Gaussian mixture model and treated by variance stabilization transformation. Some samples contain 25 replicates to construct high SNR reference data and calculate PSNR, SSIM, and Brillouin shift accuracy. Data is mainly saved in TIFF, CSV and TXT formats.