Lars Ruthotto


Creating tools that are useful (and also being used) is a rewarding part of my work. I am currently contributing to a handful of projects written inMatlab® and more recently I am experimenting with Julia. Below are some selected projects. You can find more of my work on Github.
Meganet example Meganet - a fresh approach to deep learning

Meganet is a deep learning framework written in Matlab®. Designed as a research tool, the package provides a growing number of state-of-the-art learning models. The primary focus of the package is to facilitate algorithmic research. The package supports GPU computing which enables solving medium-scale realistic learning problems. I am the lead developer of the package but I greatly appreciate the work of my collaborators .

jInv jInv - Julia Package for Parallel PDE Parameter Estimation

jInv is a flexible framework for PDE parameter estimation written in Julia. It gives easy access to many commonly used misfit and regularization terms, efficient numerical optimization, and a parallel and distributed computing model. We also have some examples online.

FAIR example FAIR - Flexible Algorithms for Image Registration with Jan Modersitzki.

FAIR is written in Matlab® and provides implementations of most state-of-the-art image registration algorithms. Although it is primarily designed as an academic and teaching tool, it offers great potential for developing prototypes and solving real-life problems. I am contributing an implementation of a hyperelastic regularizer that allows for very large and at the same time invertible deformations. This is a desirable feature in many applications of medical imaging.

HySCO ACID - Artefact Correction in Diffusion MRI with Siawoosh Mohamadi and others.

ACID is an academic software toolkit written in Matlab® that fully integrates into the batch system of SPM8. It offers many tools for pre-processing, evalation and registration of diffusion MRI data. See also: ACID download and ACID Wiki. I am contributing HySCO, a tool for correcting image distortions caused by field-inhomogeneities in EPI-MRI; described here.

VAMPIRE VAMPIRE - Variational Algorithm for Mass-Preserving Image REgistration with Fabian Gigengack and others.

VAMPIRE is a mass-preserving image registration approach that has been mainly used for motion correction in gated positron emission tomography (PET) of the human heart (see left). In this application, intensity modulations caused by the highly non-rigid cardiac motion are considered by means of a mass-preserving transformation model. The VAMPIRE approach has been shown to lead accurate and realistic motion estimates; click here.

KrylovMethods KrylovMethods.jl

KrylovMethods.jl is written in the emerging programming language Julia. It provides BLAS-based Julia implementations of some of the most useful Krylov subspace methods for solving linear systems.