CAMERA Workshop Tackles Tomographic Reconstruction

CAMERA—the Center for Advanced Mathematics for Energy Research Applications at Lawrence Berkeley National Laboratory—hosted a workshop November 9-11 in Berkeley Lab’s Shyh Wang Hall to highlight the state-of-the-art in tomographic reconstruction algorithms and software for synchrotrons, and to discuss future goals and collaborations.

The workshop brought together users, practitioners and developers of this software to assess the current landscape of available algorithms, investigate commonalities and differences among the various techniques and discuss a range of related topics, from required theoretical and algorithmic advancements on through to practical issues of implementation and deployment. Some 40 people attended.

CAMERA Workshop participants pose for a group photo outside Berkeley Lab's Shyh Wang Hall.

Berkeley Lab’s Dula Parkinson, the workshop organizer, described the rationale for this workshop: “Over the past few years I have worked with users of my micro-tomography beamline at the Advanced Light Source, and as I’ve visited and worked with beamline scientists at similar beamlines around the world, I’ve realized there are a huge number of different software packages and algorithms for tomographic reconstruction. This workshop was about seeing how to take the best ideas from each one and make them available to all tomography beamline users.”

The workshop featured technical talks from developers to describe their current algorithms and capabilities and their future potential, plus discussions from users of tomography software that focused on successes and unmet needs and finding common goals. Specific breakout sessions included:

• New algorithms and computing advances for synchrotron tomography
• A roundtable discussion on benchmarking approaches and metrics
• Software packages for synchrotron tomography
• Hands-on demonstrations of algorithms and software
• The user's perspective: success stories and unmet needs

Participants ended the workshop with a group discussion and assigned working groups to publish a workshop report. The current goal is to find a common approach to benchmarking and to set up a portal that presents algorithms and test data sets to users, along with guidance about which algorithms work best for different types of data. The participants enthusiastically agreed to make this an annual tradition—the next workshop is planned for November 2017 at Berkeley Lab.