Björn Kreisler

Generalised adjoint simulation of induced signals in semiconductor X-ray pixel detectors

The measurement quality of spatial and spectral information in an X-ray photon field is crucial to modern X-ray imaging. Direct converting semiconductor detectors with high spatial resolution and the ability to count single interacting photons, offer the possibility to obtain additional information about the spectral distribution of the X-ray photons. For being able to measure this quantity, the signal shape of the analog signal arriving at the pixel electronic input has to be well understood. The induced signals can be calculated using Ramos formulation. For gaining a complete understanding of the signal shape for different interaction points, the adjoint formulation can be applied to the problem, but up to now all considerations were restricted to sensor layers with undoped material. This is no limitation in the case of silicon, but for materials like CdTe, the doping changes the electric field significantly and therefore the signal shape and timing heavily depend on the material constants. The 3D simulation of the sensor layer of a direct converting semiconductor detector has been carried out with the finite element program COMSOL Multiphysics. The adjoint formulation has been corrected for the doping of the sensor layer. The signals arriving at the input of the electronics have been simulated and can be used for designing the pixel electronics of future detectors.

Monte Carlo simulation of pixelated photon counting X-ray detectors like the Medipix2 and the Medipix3 using high-Z sensor materials

The new generation of pixelated photon counting X-ray detectors like the Medipix2 and the Medipix3 does not measure energy deposition directly. Instead of this the measured observable is the number of photons, which have deposited an energy larger than a given threshold. To understand the response of these detectors we developed a new detector class to be used in the Monte-Carlo-Simulation ROSI, which is based on the C++ library LSCAT-GISMO, including the well-established EGS4 algorithms with its low energy extension LSCAT. The implementations in the detector class are the physics processes in the sensor layer, diffusion and repulsion of charge carriers during their drift and lifetime of the charge carriers, all taking into account intrinsic doping of the sensor material. The drift of charge carriers induces mirror charges at the electrodes leading to a signal even if not all electrons and holes reach the electrode. This results in partial charge collection and therefore has an impact on the energy resolution. The noise is modelled with Fano noise during the energy deposition, and several noise contributions of the pixel electronics. Further, the charge summing mode of the Medipix3 is available in the detector class. Due to the hybrid design of the Medipix detectors several combinations of ASIC and sensor are possible. With the implemented physics it is possible to simulate high-Z sensor materials like CdTe or GaAs in addition to silicon. The detector class allows an arbitrary positioning and orientation of the detector with respect to the incoming beam. Therefore it is possible to simulate complex imaging systems containing several units of pixelated photon counting X-ray detectors taking into account a detailed implementation of the processes inside each detector unit.

Lookup table-based simulation of directly-converting counting X-ray detectors for computed tomography

Medical computed tomography (CT) can benefit from directly-converting counting detectors. As yet there is little expertise with this type of detectors in commercially available clinical CT systems, a precise detector model is required for developing such a system. We introduce a lookup table-based simulation of counting detectors on X-ray photon level that allows studying the influence of detector parameters and efficiently evaluating proposed designs. It uses energy-resolved sinograms of incoming X-ray photons as input data and generates photon counts for each channel and reading. The effects of Poisson noise, photon interactions, pulse generation, read-out electronics and electrode signal processing are covered. The photon interaction data as well as signal characteristics are provided in the form of detector-specific lookup tables. This approach offers the precision of Monte-Carlo simulations and the efficiency of model-based descriptions. Unlike standard Monte-Carlo simulations, it is capable of simulating whole CT-scans in a reasonable amount of time on a standard workstation. Due to this efficiency, the influence of detector parameters on image quality in the reconstructed image domain can be evaluated. The simulation is verified against measured data.

Induced signals in X-ray detectors with steering grid geometry

Oncoming pixelated X-ray detectors with semiconductor sensor layers made of CdTe or GaAs can be operated in single photon counting mode. Each charge cloud generated by an interacting photon induces a signal in the electronics which is amplified, shaped and compared to a threshold. This process needs to be very fast for high flux applications, which leads to the necessity of understanding the signal induction process. The adjoint simulation of the shape and timing of the signals can be done with the finite element software COMSOL using the convection and diffusion package. For a given sensor material, the signals depend on the geometry of the electrodes and this offers ways to optimise the size and shape of the electrodes. Even additional steering electrodes can be implemented which will be the main part of this contribution. Fully 3D simulations have been carried out to study the possible effects of steering electrodes on the signal shape.

3D simulation of induced signals in the Medipix detector

The charge distribution generated by an interacting X-ray photon in the sensorlayer of an direct converting photon counting detector (as the Medipix) is transported to the pixelated electrodes and integrated during a short time. If the collected charge exceeds a threshold, a counter is increased. The generated charge will diffuse into the neighbouring pixels as the distance from the X-ray photon interaction point gets closer to a pixel edge. But even if all the charge can be projected onto a single pixel electrode, the neighbouring pixel electrodes will measure an induced signal due to the moving charge in the sensor. For being able to add these induced signals into our simulation of the whole detector device, an exact knowledge of the induced signals is necessary. Ramos formulation provides the required formula and so it is possible to calculate the induced signals for a moving charge cloud. A fully 3D simulation was performed with the finite element program COMSOL. To get a good sampling of the detector and to save computing time, the adjoint equation was solved. The calculations were carried out for better understanding of the physical processes and the imaging properties of Medipix2 and can be projected to the development for the Medipix3.

Simulation of a medical linac with evaluation of dose profiles behind an electron applicator

Cancer therapy with electrons requires a well defined electron beam. Especially a flat dose profile in the tissue is important for the treatment planning. The normal, healthy tissue should be harmed as little as possible. Therefore the understanding of the physical properties behind the linear accelerator is of outstanding importance. A simulation provides an elegant access to the track of the electrons and gamma rays towards the patient. The simulation was performed with the Monte Carlo simulation tool Geant4. The setup of a real apparatus was implemented with multiple regions for the evaluation of doses and spectral analysis. The results of the simulation provide the possibility to test new designs before prototyping.