Surrogate distributed radiological sources III: quantitative distributed source reconstructions

TL;DR

The results confirm the utility of point source arrays as surrogates for truly distributed radiological sources, and advance the quantitative capabilities of Scene Data Fusion gamma-ray imaging methods.

Abstract

In this third part of a multi-paper series, we present quantitative image reconstruction results from aerial measurements of eight different surrogate distributed gamma-ray sources on flat terrain. We show that our quantitative imaging methods can accurately reconstruct the expected shapes, and, after appropriate calibration, the absolute activity of the distributed sources. We conduct several studies of imaging performance versus various measurement and reconstruction parameters, including detector altitude and raster pass spacing, data and modeling fidelity, and regularization type and strength. The imaging quality performance is quantified using various quantitative image quality metrics. Our results confirm the utility of point source arrays as surrogates for truly distributed radiological sources, and advance the quantitative capabilities of Scene Data Fusion gamma-ray imaging methods.

Figures (14)

• Figure 1: Left: top-down view of the (un-leveled) square source reference point cloud in CloudCompare, with points colorized by height. The larger yellow point clusters are traffic cones around the source boundary and at the $10 \times 10$ grid center, while the grid of smaller yellow point clusters corresponds to the individual source locations. Right: $xy$ positions of the square source locations discernible in the (leveled) reference point cloud (black $+$) after least-squares alignment to their corresponding points in the designed square source (blue $\circ$). The sizes of the markers are intended only to best show the point-to-point alignment rather than represent any uncertainties.
• Figure 2: $35$ lidar SLAM point clouds from the WSU aerial measurement campaign, all co-registered and aligned to the idealized field coordinate system of Part I. Top: $xy$ projection. The darker spots in the center-right of the field are traffic cones placed around the boundary of the $10\times 10$ square and $\mathsf{L}$-shape sources. The circular patterns near $(x,y)=(20,10)$ and $(70,75)$ m are lidar artifacts from takeoff and landing. Middle: $xz$ projection within the $y$ boundaries of the field. Bottom: $yz$ projection within the $x$ boundaries of the field.
• Figure 3: Imaging study for four surrogate distributed sources (left to right: the $10 \times 10$ square, the $\mathsf{L}$-shape, the $8$ m separation, and the $12$ m separation sources) using the NG-LAMP system. Top row: ground truth source distributions interpolated to the $1 \times 1$ m-pitch reconstruction grid. Middle row: MAP-EM reconstructed source distributions, array source distributions, and detector trajectories derived from SLAM. Bottom row: differences between the reconstructed and interpolated true distributions, normalized by the latter. Gray pixels correspond to true source activities of $0$. The colorbar on the left corresponds to the source distributions (top row) and the reconstructed activities (middle row) while the right-hand colorbar corresponds to the bottom row relative difference maps.
• Figure 4: Like Fig. \ref{['fig:imaging_study_low']}, but for the other four source distributions, each of which featured areas of heightened source activity. Left to right: the plume, the hot/coldspot, the linear gradient, and the hot line sources.
• Figure 5: Top: MAP-EM-reconstructed (black) and measured (orange) counts in the square source study of Fig. \ref{['fig:imaging_study_low']}. Bottom: difference between the reconstructed and measured counts in terms of the deviance residual.