Review of Gadgetron

Gadgetron is an open source framework for medical imaging reconstruction. Streaming data processing pipelines are achieved through a set of modular Gadgets. There are many useful toolboxes/libraries readily available to support general image reconstruction needs, such as Cartesian and non-Cartesian parallel imaging reconstruction. Gadgetron also enables fast computing using GPUs and multi-core CPUs. Overall, I find this software a nicely-designed and user-friendly tool for MR researchers.

The modular design using different Gadgets to achieve core functionality is easy to follow. Reconstruction programs are defined by XML files, where each Gadget and its corresponding functionality is easily recognized. While most Gadgets are written in C++, the prototyping process can also be achieved using Python scripts.

Gadgetron supports the vendor-independent ISMRM Raw Data (ISMRMRD) format. Converters from vendor-specific raw data formats (Siemens, GE, Philips, Bruker, etc) to ISMRMRD are available at https://github.com/ismrmrd. The Docker image in the tutorial demonstrates an easy conversion from the Siemens data format to ISMRMRD.

The Gadgetron software is shared through GitHub and easy to install. Potential users may need to get familiar with the Docker platform and the Hierarchical Data Format (HDF5) to go through the tutorial and use the client server model. Once the streaming framework is properly set up, it is very simple to use command lines to achieve different reconstruction tasks.

With no prior experience with Gadgetron, this reviewer managed to easily follow through the tutorial without any problem. More details of Gadgetron are described in the manuscript Gadgetron: An Open Source Framework for Medical Image Reconstruction. Magn Reson Med 2013; 69:1768-1776.

In summary, Gadgetron is a well-developed open source framework for image reconstruction. I highly recommend it to MR researchers who are looking for free software to solve their general reconstruction problems and a platform to contribute new algorithms/methods to the community.

Reviewed April 2016