Gigavoxel-Scale Multiple-Scattering-Aware Lensless Holotomography - Dataset

Dataset (MATLAB files) of holograms used in: Mikołaj Rogalski, Julianna Winnik, Julia Dudek, Piotr Arcab, Emilia Wdowiak, Paweł Matryba, Marzena Stefaniuk, Piotr Zdańkowski, Maciej Trusiak,“Gigavoxel‑Scale Multiple‑Scattering‑Aware Lensless Holotomography”, submitted 2025. [2508.00567] Gigavoxel-Scale Multiple-Scattering-Aware Lensless Holotomography Contents For each of the four samples – Colosseum, Inclined_phase_resolution_target, Mouse_brain_section_500um and Multilayer_3D_phase_test – the following files are provided: Off‑axis holograms stacks<name_of_Sample>_holograms_<lambda>.mat For Colosseum, Inclined_phase_resolution_target and Multilayer_3D_phase_test, <lambda> corresponds to λ = 540 nm (green illumination) and λ = 640 nm (red illumination). For Mouse_brain_section_500um, <lambda> corresponds to λ = 405 nm (blue illumination) and λ = 561 nm (green illumination). Each file contains a three-dimensional array with dimensions [Y, X, 180], where Y and X denote the image dimensions of an individual hologram, and the third dimension indexes 180 angular projections acquired every 2° across a full 360° sweep of the illumination beam. On‑axis calibration hologram<name_of_Sample>_in_line.mat (on‑axis illumination, λ = 540 nm or 405 nm) A single on‑axis hologram used exclusively for calibration purposes. Experimental details Holograms were acquired in a custom lensless holotomography setup. The system included a board-level CMOS camera (ALVIUM 1800 U‑2050 m, 2.4 µm pixel size), with the sample placed approximately 2 mm above the sensor surface. Illumination was provided by a Supercontinuum laser (NKT Photonics SuperK), coupled into a single-mode optical fibre (Thorlabs P1‑S405‑FC‑2, ~3 µm core diameter). The output fibre was mounted on a motorized rotation stage, allowing full 360° rotation of the illumination beam in 2° steps. The azimuthal angle was set between 35° and 40°. For each of the 180 angular positions, two off-axis Gabor holograms were captured using λ = 540 nm and λ = 640 nm light, yielding 360 images per dataset. For Mouse_brain_section_500um, the corresponding wavelengths were λ = 405 nm and λ = 561 nm (due to the unavailability of the same laser source used for the other samples). Additionally, a single on-axis hologram at λ = 540 nm was acquired for calibration purposes and excluded from the tomographic reconstruction; for Mouse_brain_section_500um, the corresponding on-axis hologram was recorded at λ = 405 nm. More information on calibration and the reconstruction procedure can be found in the referenced article. Additional reconstruction/alignment parameters (provided as separate files): In addition to the hologram files listed above, the repository also contains four per-sample parameter files: <name_of_Sample>_parameters.mat Each <name_of_Sample>_parameters.mat file contains the parameters used for reconstruction and alignment: best_z — camera-to-sample distance in µm. The sign follows our propagation convention (negative values correspond to back-propagation from the sensor plane towards the sample). alpha_x, alpha_y — illumination angles in radians. Shift_X, Shift_Y — lateral shifts in pixels, used to compensate residual transverse displacements (e.g., refraction effects at the camera cover glass and small inaccuracies in angle estimation). How the parameters are intended to be used: Propagating the on-axis calibration hologram (the <name_of_Sample>_in_line.mat file for a given sample) by best_z yields an in-focus calibration object within the field of view. Propagating each off-axis hologram from the corresponding wavelength stack for that sample (e.g., 540 nm or 640 nm; for the mouse brain dataset, 405 nm or 561 nm was used during acquisition) by the same best_z using the corresponding (alpha_x, alpha_y) yields the same calibration object in focus, but typically laterally shifted. Applying the inverse pixel shifts (Shift_X, Shift_Y) aligns the reconstructions. After applying these shifts, the reconstructions are registered to the first off-axis projection, i.e., to the reconstruction obtained from the 2° hologram. Relation to figures in the manuscript The provided datasets correspond to the reconstructions shown in the paper as follows: the inclined phase resolution target is used in Figs. 2–3 (system verification and comparison of reconstruction strategies), the multilayer 3D phase test sample corresponds to Fig. 4, the 500 µm optically cleared mouse brain section corresponds to Fig. 5, and the Colosseum dataset is presented in Fig. S2 (Supplementary Information).