Compton cameras constitute a class of γ-imagers which utilize scattering processes within energy resolved particle detectors to determine 3D distributions of radionuclides. Not requiring heavy collimators as well as featuring a very large field of view, these devices have numerous potential applications in various fields of research. This includes medical imaging, homeland security and nuclear decommissioning. Though based on a well established concept, practical implementations of Compton cameras have become feasible only with most recent developments in detector technology.
In particular modern hybrid pixel detector ASICs (application-specific integrated circuits), allowing for particle detection with simultaneous high granularity energy and timing resolution, constitute promising tools in Compton imaging. Whereas the energy resolution is crucial for accurate source reconstruction, the timing resolution allows to assign coincidence accurately under high flux conditions, but also permits for advanced 3D localization of interaction points. Successful implementations of Compton cameras based on e.g. the Timepix3 ASIC have been presented recently.
Contemplating the overall performance of a Compton camera implementation, a variety of parameters need to be taken into account. This includes the ASIC/sensor/readout electronics characteristics, the device geometry, the composition of expected γ-fields, as well as the characteristics of the volumetric reconstruction methodology employed.
In order to devise designs of such cameras, specifically tailored to certain application fields, a toolbox, allowing for comprehensive system modelling and multivariate optimization of Compton cameras, is of great interest.
In the current contribution we propose such a multivariate optimization toolbox, allowing to optimize camera geometry, as well as the choice of sensors and operational parameters. Further, a pluggable design will permit for the evaluation of multiple ASICs and volumetric reconstruction algorithms in a single optimization. The optimization process will be driven by automatized evaluation of common image metrics (e.g. point spread function, signal & contrast to noise ratios), allowing for an end-to-end optimization of the full imaging chain.
The calibration of this routine against actual hardware characteristics requires to interface the toolbox against a set of benchmark measurements. As a proof of concept, such benchmark measurements will be devised and implemented for recent generations of the Timepix ASIC family.
The optimization toolbox in parts will be based on methods previously developed within or in collaboration with the Institute of Experimental and Applied Physics of CTU Prague.
This includes hardware and software for hybrid pixel detector readout, pipelines for data-handling and volume reconstruction, as well as tools for comprehensive Monte Carlo simulation. To render complex optimization processes tractable in practical use cases with available computational hardware, the latest approaches in parallel computing will be employed.
In order to verify the performance of the optimization toolbox, two practical use cases will be contemplated in a first step.
The first optimization will be towards 99mTc imaging, which is the most common radionuclide employed in pre-clinical / clinical single photon emission tomography. Exhibiting a dominant emission line at 140keV (~99%), imaging of this radionuclide does not require complex routines for spectrum analysis. However, due to the relatively low γ-energy, scatter processes in forward direction provide little energy transfer, entailing a large angular error in source reconstruction. For this reason, a camera geometry designed to maximize the capture of back-scattered photons is expected to be favorable.
With regard to our involvement in the MICADO project for nuclear decommissioning (HORIZON 2020, grant no. 847641), as a second use case the overall performance of a Compton imager, aimed towards a broad range of radionuclides to be commonly found in radioactive waste, will be optimized. Other than adequate routines for isotope identification, this advanced use case is likely to require a more complex camera design, combining multiple functional units.