OMPI Seminar: Rolf Clackdoyle and Nathan Murtha

Printer-friendly version
Thursday, October 18, 2018

Time: 3:30 - 5:00 pm, Refreshments start at 3:15 pm.

Location: Room RPB 205 (boardroom), Health Canada, 775 Brookfield Road, Ottawa. Please check in at the front desk.


1) "Tomographic Imaging of a Source in a Restricted Area Using Compton Gamma Imaging"

by Nathan Murtha, PhD Candidate, Carleton University

Supervisors: Laurel Sinclair (Natural Resources Canada) and Patrick Saull (National Research Council)

Abstract: Compton gamma imaging uses the kinematics of Compton scattering to constrain the reconstructed origin of a detected gamma ray to somewhere on a conical surface. Two-dimensional projections of these conical surfaces then form rings, the intersections of which may be used to form contours delineating where in the field-of-view the source is likely to lie. Reconstruction of the shape of a distributed source in an area restricted to entry is a particularly challenging problem which may be resolved by observing the source from multiple points of view. In March 2018, an L-shaped extended source of La-140 measuring 120 m by 20 m on one arm and 60 m by 10 m on the other arm was laid out at an experimental proving ground in Suffield, AB. This L-shaped distribution was contained in a 500 m by 500 m exclusion zone, restricting the proximity within which measurements were possible. Data were accumulated at seven positions around the exclusion zone. Using a simple back-projection method, we demonstrate preliminary results of a tomographic reconstruction of the activity distribution inside the exclusion zone.  

2) "Radar and Scatter Imaging: Dual Image Reconstruction Problems"

by Rolf Clackdoyle, L'Université Grenoble Alpes

AbstractIn synthetic aperture radar (SAR) a nominal image reconstruction problem arises. The idea is that an airplane flying a straight line along (say) the x-axis receives radar signals that are averages over circles (of various radii) centered on the x-axis. The "tomographic" reconstruction problem is to map out the radar signal at each point on the ground, based on these "circular" ray sums. This problem has been well-studied and has all the same mathematical features as the usual tomographic reconstruction problem that we are familiar with in medical imaging – there is a back-projection step which alone gives a blurred image, and pre-back-projection filtering – yielding the well-known Filtered Back-Projection (FBP) algorithm. Previous work has been presented (OMPI) that describes a system for imaging scattered radiation – a collection of overlapping "radial profiles" is obtained where each radial profile can (potentially) provide diagnostic information about a point of the sample that was irradiated with a primary beam of x-rays. In the hypothetical situation where a large collection of overlapping radial functions are acquired, all with their centers along a straight line, we see some similarity with the SAR problem, except that in this case, the unknown quantities are the individual radial functions. In this presentation, it will be shown that the two image reconstructions problems are mathematically dual to each other. We can take existing knowledge about the SAR reconstruction problem to learn about extracting radial profiles in the scatter problem.