arXiv subjects

Thomas D. MacDonald

Publications and source records attributed to Thomas D. MacDonald.

4 recordsLinked to original sources

arXiv2026

Machine learning methods for spectroscopic information recovery under ultrafast photon pileup

We present methods for recovering spectroscopic information from multiple concurrent photon interactions that would normally be lost due to pulse pileup. In particular, we focus on machine learning methods to recover information based on spatial (rather than temporal) energy deposition patterns in position-sensitive detectors. We construct two representative problems, namely (1) recovering the fraction of total energy deposition stemming from a monoenergetic signal vs. a smooth background; and (2) recovering the signal multiplicity, i.e., the number of interacting photons, in a pure-source-term example. In the signal fraction recovery problem, we use 3D convolutional neural networks (CNNs), fully-connected neural networks (FCNNs), a network based on the PointNet++ architecture, and two non-machine-learning methods to estimate the signal fraction in synthetic data when up to 20 total piled-up photons are present. The CNN, FCNN, and PointNet++ models reconstruct the signal energy deposition fractions with root mean square errors (RMSEs) of $14.5\%$, $18.8\%$, and $16.8\%$ given training datasets that fit in-core, while the classical methods perform poorly and will not improve with additional training data. In the multiplicity recovery problem, we demonstrate that, when trained with synthetically-piled-up real Cs-137 data, the 3D CNN architecture can recover the multiplicity with sub-photon RMSE, outperforming non-ML baselines. These methods can be adapted to future, more specific photon active interrogation applications, helping to re-enable spectroscopic analyses in those domains.

physics.ins-det↗

arXiv2026

Transferability of data-driven optimization results across multiple pixelated CdZnTe spectrometers

Recent work by Vavrek et al. (2025) showed that machine learning methods can be used to exploit spatial patterns of performance variations within the highly-segmented H3D M400 gamma spectrometer to improve an overall spectroscopic performance metric. That work also introduced the spectre-ml software, which tests various greedy, heuristic, random, and machine learning clustering algorithms to find the best performing mask for excluding detector regions to improve a user-defined performance metric by training on a given dataset. In this work, we build off of Vavrek et al. (2025) and seek to determine to what extent an optimized binary voxel mask trained on a given dataset can generalize to other datasets. In particular, this paper evaluates the transferability of masks trained on one M400 dataset to another M400 detector, in order to determine whether the total effort required in designing masks for different detectors and applications can be substantially reduced by using a single common mask. It also examines testing and training on different subsets of the same dataset to determine the natural level of variability in optimization results. In the inter-detector analysis, as expected, the best performing model on each detector is often one trained on that dataset, with an average performance enhancement of $16\%$ when considering the relative uncertainty in a Doniach fit to the $186$ keV peak. In comparison, the best transferred masks, with the best on average performance metric across all six detectors, show only a slightly smaller improvement of $13\%$ on average. These results suggest that high-performing, well-transferable masks can be shared among detectors, reducing or even eliminating the laborious processes of collecting a training dataset and performing the optimization for each detector, ultimately improving safeguards efficiency.

physics.ins-det↗

arXiv2025

Data-driven optimization of pixelated CdZnTe spectrometers for uranium enrichment assay

In recent work [Vavrek et al. (2025)], we developed the performance optimization framework spectre-ml for gamma spectrometers with variable performance across many readout channels. The framework uses non-negative matrix factorization (NMF) and clustering to learn groups of similarly-performing channels and sweep through various learned channel combinations to optimize the performance tradeoff of including worse-performing channels for better total efficiency. In this work, we integrate the pyGEM uranium enrichment assay code with our spectre-ml framework, and show that the U-235 enrichment relative uncertainty can be directly used as an optimization target. We find that this optimization reduces relative uncertainties after a 30-minute measurement by an average of 20%, as tested on six different H3D M400 CdZnTe spectrometers, which can significantly improve uranium non-destructive assay measurement times in nuclear safeguards contexts. Additionally, this work demonstrates that the spectre-ml optimization framework can accommodate arbitrary end-user spectroscopic analysis code and performance metrics, enabling future optimizations for complex Pu spectra.

physics.ins-det↗

arXiv2018

Experimental Demonstration of Multiple Monoenergetic Gamma Radiography for Effective Atomic Number Identification in Cargo Inspection

The smuggling of special nuclear materials (SNM) through international borders could enable nuclear terrorism and constitutes a significant threat to global security. This paper presents the experimental demonstration of a novel radiographic technique for quantitatively reconstructing the density and type of material present in commercial cargo containers, as a means of detecting such threats. Unlike traditional techniques which use sources of bremsstrahlung photons with a continuous distribution of energies, multiple monoenergetic gamma radiography (MMGR) utilizes monoenergetic photons from nuclear reactions, specifically the 4.4 and 15.1 MeV photons from the $^{11}$B(d,n$γ$)$^{12}$C reaction. By exploiting the $Z$-dependence of the photon interaction cross sections at these two specific energies it is possible to simultaneously determine the areal density and the effective atomic number as a function of location for a 2D projection of a scanned object. The additional information gleaned from using and detecting photons of specific energies for radiography substantially increases the resolving power between different materials. This paper presents results from the imaging of mock cargo materials ranging from $Z\approx5$--$92$, demonstrating accurate reconstruction of the effective atomic number and areal density of the materials over the full range. In particular, the system is capable of distinguishing pure materials with $Z\gtrsim70$, such as lead and uranium --- a critical requirement of a system designed to detect SNM. This methodology could be used to screen commercial cargoes with high material specificity, to distinguish most benign materials from SNM, such as uranium and plutonium.

physics.ins-det↗