OS1  Software Focused on Image Reconstruction

Saturday, Nov. 7  13:00-16:30  Pacific Salon 3

Session Chair:  Kris Thielemans, University College London, United Kingdom; Charalampos Tsoumpas, University of Leeds, United Kingdom

Show/Hide All Abstracts

(13:00) OS1-1, Open Source Software Focused on Image Reconstruction

C. Tsoumpas1, K. Thielemans2

1Div. Biomedical Imaging, University of Leeds, Leeds, United Kingdom
2Institute of Nuclear Medicine, University College London, London, United Kingdom

Open source software is established as one of the new successful ways of accelerating software development and scientific discoveries across different scientific domains. In tomography, several packages have been developed and extensively used. The aim of this workshop is to provide a summary of existing packages in tomographic image reconstruction relevant to the conference attendees. We will introduce the invited speakers who will present with sufficient technical detail the most recent software and algorithmic developments on open access reconstruction software libraries for PET, SPECT, CT and MR.

(13:05) OS1-2, STIR: Software for Tomographic Image Reconstruction

K. Thielemans1,2, C. Tsoumpas3

1Institute of Nuclear Medicine, University College London, London, UK
2Algorithms & Software Consulting, London, UK
3Biomedical Imaging, University of Leeds, Leeds, UK

STIR is an open source object-oriented library for tomographic image reconstruction implemented in C++ for 3D positron and single photon emission tomography. This library has been designed such that it can be used for many algorithms and scanner geometries, while being portable to various computing platforms. The current release includes several features such as motion compensated image reconstruction, scatter and randoms estimation and correction, additional iterative reconstruction algorithm, parametric image reconstruction and many other tools. STIR offers to both academia and industry an environment for thoroughly validated software of advanced image reconstruction techniques and a means for faster translation of newly developed algorithms in preclinical and clinical research and practice. This presentation will elaborate on the advantages and complexities of the development and maintenance of such a large open source library for primarily research purposes and will discuss potential strategies for future directions.

(13:35) OS1-3, Occiput: Unified PET, MR and SPECT Reconstruction

S. Pedemonte1,2, N. Fuin1, S. Bowen1, C. Catana1, K. Van Leemput1,2,3

1Athinoula A. Martinos Center for Biomedical Imaging, MGH & Harvard, Boston, MA, USA
2Dept. of Computer Science & Dept. of Neuroscience & Biomedical Engineering, Aalto University, Espoo, Finland
3Dept. of Applied Mathematics & Computer Science, Technical University of Denmark, Kongens Lyngby, Denmark

Occiput is an open source GPU-based reconstruction software for PET, MR and SPECT inspired by principles of artificial intelligence and designed for dynamic and multi-modal imaging. We describe the features and the design of the software and present live demos of applications in PET and PET-MR.

(14:05) OS1-4, CASTOR: Customizable and Advanced Software for Tomographic Reconstruction

T. Merlin1, S. Stute2, C. Comtat2, F. Lamare3, D. Benoit1, J. Bert1, A. Autret1, D. Visvikis1

1INSERM U1101, LaTIM, National Institute of Health & Clinical Sciences, Brest, France
2Imagerie Moléculaire In Vivo, Institut d'Imagerie Biomédicale, Orsay, France
3Dept of Nuclear Medicine, University Hospital Bordeaux, Bordeaux, France

Our aim is to develop and offer to the scientific community an open-source toolkit for multidimensional multimodality tomographic reconstruction. GPU image reconstruction methodologies will be integrated in the software along-side a parallel CPU implementation. The proposed platform is aimed as a flexible tool providing basic image reconstruction features for average users, but also allowing advanced users to incorporate their own methodologies considering the different parts of an image reconstruction algorithm (such as specific projectors, optimizers, different data corrections, kinetic modeling) through the implementation of new classes. The main challenge for this project is to design a generic core framework, flexible enough to cope with the specificities of each modality but also provide a user-friendly class system for the incorporation of new methodologies.

(14:35) OS1-5, A Python Library For Inverse Problems

J. Adler1,2, O. Öktem2, H. Kohr2

1Elekta, Stockholm, Sweden
2KTH Royal Institute of Technology, Stockholm, Sweden

Operator Discretization Library (ODL), a library currently under development for use in inverse problems such as Computed Tomography (CT) and Single Photon Emission Computed Tomography (SPECT) is introduced. The library is written in Python and is intended to allow researchers to write prototypes suitable for real life data by exploiting existing hardware accelerated forward and back projection codes in a uniform manner using the language of mathematical analysis. The library has two parts. The analysis part allows users to define mathematical objects such as linear operators and vector spaces as well as methods on these such as the conjugate gradient method. The tomography part allows the definition of acquisition geometries in continuum and the discretization of these. We also plan to provide wrappers for some widely used libraries such as ASTRA, STIR, and others. We present some examples and initial comparisons of the library with existing libraries suggesting that reconstruction algorithms written using the library has comparative speed with fully optimized C++, while allowing a much clearer and concise syntax and simple access to the scientific computing tools of the Python language.

(15:05) OS1-6, The ASTRA Toolbox: Fast and Flexible Tomographic Reconstructions

W. van Aarle1, W.-J. Palenstijn2, J. Sijbers1

1Dept. of Physics, University of Antwerp, Antwerp, Belgium
2Scientific Computing, Centrum Wiskunde Informatica, Amsterdam, The Netherlands

We present the ASTRA Toolbox as an open platform for 3D image reconstruction in tomography. The ASTRA Toolbox is a software platform developed at the University of Antwerp, Belgium, and at the Centrum Wiskunde Informatica (CWI), Amsterdam, The Netherlands, to address the need for a fast, flexible development platform for tomography algorithms. It provides a set of building blocks that can deal with various geometrical setups and incorporate a variety of constraints in an efficient manner. The toolbox is accessible through MATLAB and Python, providing a powerful platform for algorithm prototyping, and is available as open source software under a GPLv3 license. Due to its flexible nature, the ASTRA Toolbox is suitable for addressing a wide range of computational problems in many tomographic applications such as medical CT, biomedical or industrial micro-CT, synchrotron tomography, and electron tomography. It offers full 3D flexibility for modelling special geometric setups and allows to perform parallelized computations using such complex geometrical setups on Graphics Processing Units (GPUs). Through its integration in MATLAB and Python, advanced numerical code such as regularized reconstruction algorithms (e.g. total variation minimization (TV-min)), can be directly applied to large experimental datasets.

(15:35) OS1-7, Gadgetron: An Open Source Framework for Medical Image Reconstruction

M. S. Hansen

National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, USA

The Gadgetron is an open source medical image reconstruction framework. It has primarily been used for MRI reconstruction. The framework implements a flexible system for creating streaming data processing pipelines where data pass through a series of modules or “Gadgets” from raw data to reconstructed images. The data processing pipeline is configured dynamically at run-time based on an extensible markup language configuration description. The framework promotes reuse and sharing of reconstruction modules and new Gadgets can be added to the Gadgetron framework through a plugin-like architecture without recompiling the basic framework infrastructure. Gadgets are typically implemented in C/C++, but the framework includes wrapper Gadgets that allow the user to implement new modules in the Python or Matlab for rapid prototyping. In addition to the streaming framework infrastructure, the Gadgetron comes with a set of dedicated toolboxes in shared libraries for medical image reconstruction. This includes generic toolboxes for data-parallel (e.g., GPU-based) execution of compute-intensive components.

(16:05)  Discussion