Open-source Software


elsa is a modern, elegant C++ library intended for use in tomographic reconstruction. Currently, operators are implemented for the forward model of X-ray Computed Tomography (the X-ray transform). Other imaging modalities can be supported by implementing appropriate operators.


curadon aims to be a foundational library designed to unite model-based and deep learning communities in the X-ray CT field. As a versatile building block for a broad range of X-ray CT applications, curadon offers efficient projection operators, a minimalistic interface, easy-to-use Python bindings, and support for diverse geometries. Demonstrating both efficiency compared to standard tools and flexibility with real-world data, curadon caters to classical analytic and model-based iterative approaches, as well as deep learning applications.

Robotic sample holder

Our software package for a robotic sample holder is designed to enable robotic X-ray computed tomography, building on ROS. It offers support for path planning, collision detection, calibration, and generation of arbitrary acquisition trajectories.


CWFA contains code for a conditional wavelet flow architecture for 3D reconstruction of XLFM images. This code is associated with our preprint arXiv:2306.06408.


WindowNet contains code for learnable windows for chest X-ray classficiation. This code is associated with our publication WindowNet: Learnable Windows for Chest X-ray Classification.


FusionM4Net contains code to perform multi-modal multi-label skin lesion classification. This code is associated with our publication FusionM4Net: A multi-stage multi-modal learning algorithm for multi-label skin lesion classification.


SLNet / XLFMNet contains code to perform real-time light field 3D microscopy via sparsity-driven learned deconvolution. This code is associated with our publication Real-Time Light Field 3D Microscopy via Sparsity-Driven Learned Deconvolution.


LFMNet contains code to reconstruct a 3D confocal volume from a 4D light field image. This code is associated with our publication Learning to Reconstruct Confocal Microscopy Stacks From Single Light Field Images.


oLaF is a flexible and efficient Matlab framework for 3D reconstruction of light field microscopy data. It is designed to cope with various LFM configurations in terms of MLA type (regular vs. hexagonal grid, single-focus vs. mixed multi-focus lenslets) and placement in the optical path (original 1.0 vs. defocused 2.0 vs. Fourier LFM designs).