Results
A.
Physical characterisation:
General comparison between CR and DR systems and between direct and indirect conversion technology in DR systems: Figures 1,
2 and 3 show for several systems the frequency dependence of the parameters MTF,
NNPS and DQE,
up to the Nyquist frequency.
- DR units clearly outperformed CR systems.
- Direct conversion flat panels (a-Se,
systems A1-A4) exhibited a higher MTF than detectors based on indirect conversion with needle-like CsI (systems B1,
B2) (Fig.
1).
- Flat panels using direct conversion technology were generally affected by a higher noise than those using indirect conversion (Fig.
2).
- Considering DQE,
units with direct conversion technology resulted superior (peak DQE > 65% versus ca.
50% for units with indirect conversion) (Fig.
3).
Comparison between different a-Se readout technologies: System A3 uses conventional a-Si TFT-matrix readout,
System A4 an innovative photoconducting switching readout obtained by a second a-Se layer,
separated by a thin electron trapping layer (ETL) [13].
- Although the detector of System A4 has a smaller pixel pitch,
its MTF value was lower compared to System A3 (Fig.
4).
- The detector with photoconductive switching readout exhibited low noise,
in particular at higher frequencies,
thus overcoming the high-noise-problem of a-Se detectors with conventional TFT-matrix (Fig.
5).
- Peak DQE values were comparable in Systems A3 and A4,
above 70%; System A4 with the photoconductive switching readout achieved a better overall DQE performance having a slower decrease (Fig.
6).
B.
Image quality assessment:
- Exposure technique optimization: In all systems target/filter combinations producing an x-ray spectrum of higher mean energy showed up to be more dose-efficient.
For a phantom thickness ≥ 3 cm PMMA the figure of merit IQFinv2/AGD was higher using Mo/Rh,
Rh/Rh or W/Rh than with Mo/Mo.
Fig.
7 reports an example.
Systems A3 and A4 resulted already optimized with the factory AEC-settings.
- Comparison between systems with indirect and direct conversion AMFPI in optimized exposure technique: The studied direct conversion AMFPIs scored a slightly higher IQFinv2/AGD value than indirect conversion one; the difference is significant (Wilcoxon test,
p < 5%).
- Comparison between conventional direct AMFPI technology and innovative photoconducting switching readout: System A4 with the photoconductive switching readout scored the highest IQFinv2/AGD values among all direct conversion flat panel systems at every phantom thickness (Wilcoxon test,
p < 1%).
Fig. 8 shows the direct comparison between Systems A3 and A4,
which differ only by the detector readout.
The higher IQFinv2/AGD value of System A4 traduces into 8%-28% dose savings in terms of AGD compared to System A3.
- Comparison of image quality with target/filter of clinical use at equal AGD: Fig.
9 shows a comparison of the IQFinv value of all systems at 1.45 mGy,
where the systems with a newer detector technology are superiore (Systems A3,
A4 and D1).
Fig.
10 compares the systems at a lower dose,
ca.
0.8 mGy AGD,
where the slit scanning System D1 excels.
C.
Evalution of different acquisition techniques:
For the newer target/filter combinations not considered in the European Guidelines [14],
the correction factors necessary for AGD calculation according to the method of Dance were kindly supplied by Dance.
- Comparison of AGD in 2D and 3D (Tomosynthesis) acquisition: The overall dose of a tomosinthesis acquisition turned out to be well below the dose of a standard screening exam (two projections),
confirming the manufacturer’s statement (-15% - -40%).
- Comparison of slit scanning/photon counting detection technology with conventional full field projection technology: At low AGD values System D1 with slit scanning acquisition and photon counting detection technology (intrinsic scatter radiation rejection) showed markedly higher IQFinv scores than the other systems with full field projection acquisition.
On the other hand,
increasing the exposure permitted only a reduced relative quality improvement.
Accordingly System D1 operates in clinical use at particularly low exposure levels.
At medium doses slit scanning and photoconductive switching readout technologies demonstrated comparable image quality.