Further Software

Several software packages developed at our institute are either suitable for both MPI and MRI or neither. These software packages are grouped in different GitHub organizations or under the account of Tobias Knopp.

The packages AbstractImageReconstruction.jl and RegularizedLeastSquares.jl contain code that is used by both MPIReco.jl and MRIReco.jl. The former contains an abstract interface for tomographic image reconstruction algorithms and code that stores and simplifies working with reconstruction algorithm parameters. The latter contains several solvers that can solve large linear systems using regularization techniques and nonlinear problem formulations.

The RedPitayaDAQServer repository contains software for use with RedPitaya's STEMlab 125-14 devices. These devices together with our software allow continuous and parallel generation and acquisition of analog signals with sampling rates up to 15.625 MH/z. In addition, multiple RedPitayas can be synchronized to form a cluster. These clusters are responsible for the analog signal handling in many of our hardware projects.

The Julia package NFFT.jl provides an implementation of the nonequidistant Fast-Fourier Transform, that is completely generic and dimension-agnostic, requiring about two to three times less code than the well-known libraries NFFT3 and FINUFFT while still being one of the fastest NFFT implementations developed to date.

The Julia package SphericalHarmonicExpansions.jl provides methods to numerically handle real spherical harmonic expansions and their coefficients. These methods together with the Julia package MPISphericalHarmonics.jl are used in our magnetic field characterization project to investigate the magnetic fields of MPI devices.

Project Publications

[191962]
Title: TrainingPhantoms.jl: Simple and Versatile Image Phantom Generation.
Written by: P. Jürß, C. Droigk, M. Boberg, and T. Knopp
in: <em>International Journal on Magnetic Particle Imaging</em>. Mar (2025).
Volume: <strong>11</strong>. Number: (1 Suppl 1),
on pages: 1-2
Chapter:
Editor:
Publisher:
Series:
Address:
Edition:
ISBN:
how published:
Organization:
School:
Institution:
Type:
DOI: https://doi.org/10.18416/IJMPI.2025.2503031
URL: https://www.journal.iwmpi.org/index.php/iwmpi/article/view/802
ARXIVID:
PMID:

[www]

Note: inproceedings, opensoftware, generalsoftware

Abstract: Large collections of labeled data play a crucial role in supervised machine learning projects. Unfortunately, such datasets are quite rare in the medical domain. In this work, the Julia project TrainingPhantoms.jl is introduced, which provides a simple interface to generate large and diverse collections of randomly generated image phantoms. The proposed phantom generator has been successfully used to train an image quality enhancement network that managed to generalize to unseen experimental out-of-distribution data.