Flat-panel detector

For the Monte Carlo simulation, two types of FPD base models were chosen: first, a model resembling the AN9 detector of our previous study (Fast et al. 2012) with a Lanex Fast-B phosphor screen and without a Cu build-up plate; second, a model closely resembling the AL7 detector, which was available for experiments in this study. The AL7 detector features a Lanex Fine phosphor screen and a 1 mm Cu build-up layer. The only physical difference between the two detectors is the phosphor screen and the Cu build-up plate. The internal composition of the detector models is summarized in table 4.1. Parameters are based on previously published values from Antonuk et al. (1990), Schach von Wittenau et al. (2002), Parent et al. (2006), Pistrui- Maximean et al. (2006), and Cho et al. (2008), as well as on our own estimates. A schematic drawing of the detector and its orientation in respect to the kV and MV beams is given in figure 4.1.

(a) Conventional geometry. (b) In-line geometry.

Figure 4.1 Comparison of (a) conventional x-ray geometry used for kV or MV imaging and (b) the in-line x-ray geometry. For geometry (b) the kV beam is tilted in respect to the MV beam to achieve a geometric separation of kV and MV fields (Fast et al. 2012). Note that the (greatly simplified) detector components are not to scale and that in reality visible light is emitted isotropically. Secondary electrons are not shown.

The Lanex screens are divided into several layers (e.g. Cho et al. (2008)) with the scintillation taking place in a layer of terbium-doped gadolinium oxysulfide Gd₂O₂S : Tb. The Tb concentra-

Table 4.1List of layers in our FPD model and their physical properties. Estimated values for the front and back layers are based on measurements taken of the AL7 detector. Note that our AN9 detector does not have the Cu build-up plate.

Layer Compounds Thickness Density [mm] [g· cm−3] Front plate Aluminium 0.85 2.7

Gap Air 5.8 0.001

Build-up Copper 1.0 8.96

Gap Air 1.0 0.001

Enclosure Graphite 0.5 2.2

Lanex Fast-B or Fine screen

Coating Acetate 0.01 1.32 Plastic support Polyethylene 0.178 1.4 Reflective surface TiO₂ n/a n/a Phosphor Gd₂O₂S : Tb 0.36 or 0.09 3.72 Coating Acetate 0.008 1.32

Sensor a-Si 0.001 2.33

Glass substrate Corning1737 1.1 2.54

Supporting material

Support Carbon fibre 3.0 1.6 Support Aluminium 2.0 2.7

Gap Air 15.1 0.001

Printed-Circuit Board

Circuits Copper 0.1 8.96 Substrate Epoxy, glass 3.5 1.9

Gap Air 6.8 0.001

Back plate Aluminium 3.0 2.7

tion is usually smaller than 0.1% and has thus little effect on the x-ray absorption. The phosphor layers have areal densities of 34 mg· cm−2 (Lanex Fine) and 133 mg· cm−2 (Lanex Fast-B) re- spectively (El-Mohri et al. 2001). Considering that the phosphor layer is in reality not pure Gd₂O₂S : Tb, but a mixture with a polymer binder and air pockets, it is reasonable to assume an effective phosphor packing density of 50% (Liaparinos et al. 2006). This results in a Gd₂O₂S : Tb mass density of 3.72 g· cm−3 and a thickness of 90 μm (Lanex Fine) and 360 μm (Lanex Fast-B).

4.3.1.1 Detector configurations.

Starting from our previously described model of the AL7 detector, we have removed selected layers of the FPD to investigate the impact on the imaging performance. Reducing the number of layers and thus the absorption length is the common theme behind most of our derived non-standard detector configurations. Making the detector thinner should not only reduce the absorption of the MV beam, it should also reduce scatter within the panel and scatter onto the patient. The following layer designs were studied:

1. Standard AL7 detector: described in detail in table 4.1.

2. Stripped-down detector: based on the standard detector but without the Al front plate, Al back plate, and Cu build-up plate.

3. Direct detector: based on the standard detector but without the Al back plate, Lanex screen, Cu build-up plate, and graphite layer.

A Monte Carlo Study of an Improved X-Ray Detector

4. Notional detector: based on the standard detector but the Al front plate is replaced with carbon fibre, no Al back plate, no printed-circuit board at the back of the detector, slimmed down support structure.

The phosphor screen is removed for the direct detector configuration. Consequently, the measured signal only depends on the directly absorbed energy in the a-Si. Because the a-Si layer is only about 1 μm thick, this would usually mean a sharp drop in signal. For the in-line geometry, however, all FPD components that are usually ‘down-stream’ are now acting as build-up layers for the direct detection. Removing the entire phosphor screen acted as surrogate for removing the phosphor in the MV region only (experimentally this was impossible without damaging the phosphor screen). In a realistic scenario, the phosphor is required in the kV region of the detector in order to obtain good contrast.

The central support structure of the proposed notional detector, composed of carbon fibre and aluminium, is reduced to an overall thickness of 2.6 mm. The Al back cover was completely removed. We also assumed that the wires present in the printed circuit-board could in principle be rerouted around the detection area, and removed this layer as well. Additionally, the Al front cover listed in table 4.1 was replaced by carbon fibre. The rationale for choosing the given materials and thicknesses was to minimise x-ray absorption, while at the same time ensuring mechanical stability and protecting the inside of the flat-panel detector from external influences. It should be noted that depending on the detector to MV source distance, only parts of the detection area of the panel would need to be reduced to the given thickness. For example, a distance of 60 cm would require a ‘thin’ area of approximately 25x25 cm2_.