Abstract

Today, in commercial CR systems BaFBr_1-xI_x:Eu²⁺ powder phosphor screens are used as imaging detectors (1).

It has been discovered that CsBr:Eu²⁺ is an excellent storage phosphor combining a high specific X-ray absorption and the possibility of needle growth.

The relevant storage phosphor properties of a CsBr:Eu²⁺ needle screen have been measured. Its blue emission, peaking at 440 nm and near I.R. stimulation band, with maximum at 685 nm, make it well suited for use in CR systems. Sensitivity and sharpness of a 150m and a 500m thick CsBr:Eu²⁺ needle screen were measured in a flying-spotscanner. At 220kVp and 8mm of Cu filtration, for 500m CsBr:Eu²⁺ equal sharpness is obtained for 13.5 % versus 6.4 % X-ray absorption in BaFBr_1-xI_x:Eu²⁺ and for 150m much higher sharpness is obtained. The number of photostimulated light quanta per absorbed X-ray quantum is higher than for BaFBr_1-xI_x:Eu²⁺. The low stimulation energy allows read-out with a very low power stimulation source, the dark-decay is acceptable and the phosphor can be erased with a fraction of the erasure power that is used in today's commercial CR readers.

For the first time in 30 years a storage phosphor has been discovered that performs better than the "go2.2 lden standard" BaFBr_1-xI_x:Eu²⁺. Use of the CsBr:Eu²⁺ needle screen in existing CR readers and in new families of CR readers could lead to significantly improved image quality. This is demonstrated by the fact that much higher SNR is obtained for the CsBr:Eu²⁺ needle screen than for the BaFBr_1-xI_x:Eu²⁺ powder screen in the cascaded system analysis.

1. Introduction

Computed radiography (CR) was introduced as a technology for digital radiography ca. 15 years ago (1). In CR, storage phosphors are used that store energy upon X-ray exposure. A plate containing the storage phosphor is exposed in a light-tight cassette and then read out in a flying spot scanner to create the digital image. Today, CR technology is well established. The system is cheap, has good producibility and is robust.

Alternative digital radiography systems [Direct Radiography (DR)] have been developed to convert X-ray signals into electrical signals, either directly as in a-Se based systems or indirectly as in the systems using a scintillator layer in combination with a-Si photodiodes.

2. Storage phosphor image plates

2.1. Needle image plate vs. powder image plate
Conventional phosphor plates, used in film/screen radiography and the storage phosphor plates in the Agfa, Fuji and Kodak CR systems are powder plates. The active layer consists of microscopic phosphor crystals, held together by a binder.

The CsI:Tl layer in certain DR systems is an array of phosphor needles with a 5 to 10 m diameter and a 400 to 500 m length. Its morphology is similar to that of the CsI:Na scintillator layers used in conventional image intensifiers (2).

A needle plate leads to better image quality than a powder plate for two reasons. The needles act as light guides, thereby strongly reducing light spread in the phosphor layer. As a consequence, image sharpness is much higher at equal thickness. Also, no binder is present, which implies a higher phosphor packing density and, thus, a higher X-ray absorption.

2.2. CsBr:Eu²⁺ needle imaging plate (NIP)
Today, BaFBr_1-xI_x:Eu²⁺ is mainly used in commercial CR systems.

It is used in the plates of Agfa, Fuji and Kodak and now also Konica. It is an excellent storage phosphor with a high storage capacity (3) and a relatively high specific X-ray absorption. Due to its rather complex chemical composition, BaFBr_1-xI_x tends to decompose upon vapour deposition, which makes needle growth impossible.

Using high throughput synthesis and screening methodologies, Agfa, in co-operation with SYMYX, discovered an excellent new storage phosphor, namely CsBr:Eu²⁺.

Its chemical composition and density of 4.5 g/cm³ lead to a specific X-ray absorption that is similar to that of
BaFBr_1-xI_x.

The relevant spectroscopic and storage phosphor properties of CsBr:Eu²⁺ have been measured.

Image quality of a CR system with the CsBr:Eu²⁺ NIP has been determined and a comparison has been made with the image quality of a state-of-the-art CR system, with the image quality of an a-Se based flat-panel detector and with the image quality of the different NDT film system classes.

3. Experimental

3.1. CsBr:Eu²⁺ NIP preparation
A 150 m and a 500 m thick CsBr:Eu²⁺ NIP were made via thermal vapour deposition. To this aim, CsBr was mixed with EuOBr and placed in a container in the vacuum deposition chamber. The phosphor was deposited on a glass plate of 25 x 25 cm². During vapour deposition, the substrate was rotated at 12 rpm to obtain good thickness homogeneity. A small reference plate was prepared under similar conditions for measurement of the spectroscopic properties and of the conversion efficiency.

3.2 Scanning system for NIP read-out
Read-out was done in a flying spot scanner. The scanning light source was a 30 mW diode laser emitting at 680 nm. Reference MD-10 BaFBr:Eu²⁺ plates were scanned on the same scanner for sharpness measurement, for measurement of read-out depth and for image quality measurements.

Pixel size was 140 m for measurement of the modulation transfer function(MTF) and for measurement of signal-to-noise ratio (SNR). For measurement of erasability and read-out depth the pixel size was 50 m . The scanning spot had a width that matched the pixel size.

3.3. Measurement of phosphor properties
Emission spectrum measurement
Photoluminescence measurements were performed at room temperature on a SPEX DM3000F spectrofluorometer equiped with 0.22 m SPEX 1680 monochromators (resolution 0.1 nm) and a 450 W xenon lamp. The excitation UV-light had a wavelength of 370 nm. The emission spectra were measured and corrected for the photomultiplier sensitivity and for the spectrum distortion by the monochromator.

Stimulation spectrum measurement
In the stimulation spectrum measurement the photostimulated emission was measured as a function of stimulation wavelength using the following set-up:

The light of a quartz-iodine tungsten lamp was fed into a monochromator and then mechanically chopped with a rotating wheel with a single slit. The second harmonic of the monochromator was eliminated by placing a 4 mm Schott GG435 filter in front of the phosphor screen. By chopping the stimulating light (duty cycle 1/200) only a small fraction of the absorbed energy in the phosphor was released as emission light. The emission light intensity was measured using a photomultiplier tube. Only the AC signal was measured to eliminate the offset caused due to e.g. the dark current of the photomultiplier. A good signal to noise ratio was obtained by averaging several pulses. Correction was made for the wavelength dependence of the tungsten lamp emission intensity and for the transmission of the grating as a function of wavelength.

Conversion efficiency measurement
The conversion efficiency (C.E.), i.e. the total photostimulated light energy emitted after X-ray exposure was determined for CsBr:Eu²⁺ and for commercial BaFBr:Eu²⁺.

Phosphor screens were exposed with an X-ray source operating at 80 kVp and 5 mA. A 21-mm thick external Al filter was used to harden the X-ray spectrum. In the measuring set-up a 5-mW diode laser emitting at 690 nm was used to photostimulate the X-ray irradiated NIP phosphor screen and a 10-mW He-Ne laser for the BaFBr:Eu²⁺ plate.

The laser-optics comprised an electronic shutter, a beam-expander and a filter. Using a diaphragm placed in contact with the screen the light emitted by only 7 mm² was collected.

The stimulating laser light, transmitted by the screen and the stimulated emission light were separated by a 4 mm BG 39 SCHOTT (trade name) filter in the case of the CsBr:Eu²⁺ NIP and by a 4 mm BG3 SCHOTT (trade name) filter in the case of the BaFBr:Eu²⁺ screen, so that only the emitted light reached the photomultiplier. The signal intensity was measured as a function of time.

Erasability measurement
In the erasability measurement, storage phosphors were exposed to 80 kVp radiation. The extent of signal reduction by erasure with a 500 W (electric power) quartz-halogen lamp for 1 s was measured.

Measurement of required read-out power
The light energy needed to set free the stored energy upon scanning the CsBr:Eu²⁺ NIP was compared with the light energy needed to scan a BaFBr:Eu²⁺ phosphor plate.

The screens were scanned a number of times without intermediate erasure until the signal was immeasurably small. From the signal evolution with the number of read-out cycles the read-out depth as a function of read-out energy was calculated.

3.4. Image Quality Measurements
Resolution
For film the detector MTF reduces to the inherent unsharpness (U_f) i.e. the unsharpness of the incoming X-ray image, which increases with radiation energy. For screen/film systems resolution is mainly determined by the optical scatter in the phosphor layer. For CR where a storage phosphor screen is scanned with a flying spot scanner, one has to account for both the screen and the scanner MTF. Only the screen MTF is slightly dependent on the radiation energy in the energy range W:50-320 kVp.

The MTF of the system was determined from the square-wave response function (SWRF) of periodic bar patterns. The amplitude response at various spatial frequencies was analysed between 0.5 and 5 lp/mm. The SWRF was determined by normalising there amplitudes with respect to the contrast of the bar at the very low frequency of 0.025 lp/mm.

The MTF can then be calculated from the SWRF according to the equation 1:

(1)

SNR
SNR was measured for the BaFBr:Eu MD10 plate for the exposure conditions mentioned in NDT standard EN584-1 i.e. W:220kVp pre-filtered with 8mm Cu (HVL 3.2mm Cu). A composite filter (100mu Pb and 0.5mm Fe) was applied on top of the phosphor layer. The Pb layer faced the source side of the detector.

The scanner gain was set in order to achieve mid-scale digital signal values (i.e. 2000 ± 100 on 12 bit scale) using a commercially available storage phosphor screen (Agfa MD10) at an exposure of 36.8 mR in the detector plane (i.e. 10% of Structurix D7).

The scanner prototype used square root hardware compression before A/D conversion. In order to measure the SNR the IP was uniformly exposed. The intensity over dose curve was measured. The relationship between the digital signal intensity values and the radiation dose was then derived. Using this relationship the signal values were converted. Subsequently, only these linearized data are used for the noise analysis.

Then, the signal intensity I_meas and standard deviation s_PSL are calculated from 11 data sets per sample by 512 values per test area. Each set should be filtered by a high pass filter with 0.1 lp/mm and 11 s_PSLi values are calculated. The final s_PSL value is obtained as the median of all 11 s_PSLi values. This procedure complies with the European standard proposal concerning the classification of storage phosphor systems

4. Results

4.1. Phosphor properties
Stimulation and emission spectra
Figure 1 shows the emission and stimulation spectra of CsBr:Eu²⁺. The blue emission, centred around 440 nm allows efficient detection with both a photomultiplier tube and a CCD.

The stimulation band, centred around 680 nm allows efficient read-out with cheap and compact diode lasers.

Fig 1: emission and stimulation spectra of CsBr:Eu2+.

Conversion efficiency
From the intensity curve as a function of time, the total amount of photostimulated light emission energy was calculated. After all necessary corrections, a conversion efficiency value (C.E.) is obtained in pJ/mm²/mR. The obtained quantity was divided by the screen thickness to obtain the conversion efficiency per volume of phosphor, i.e. in pJ/mm³/mR. Conversion efficiencies of 29 pJ/mm³/mR for BaFBr:Eu²⁺ and of 37 pJ/mm³/mR for CsBr:Eu²⁺ were found, respectively.

Taking into account the X-ray absorption of the samples, the emission photon energies and the number of X-ray quanta per mR, this corresponds to ca. 500 PSL photons per absorbed 50 keV X-ray quantum for BaFBr:Eu²⁺ and to ca. 750 PSL photons per absorbed 50 keV X-ray quantum for CsBr:Eu²⁺.

Erasability
Under the abovementioned erasure conditions, the signal is reduced to 7.10^-5 of the original signal for CsBr:Eu²⁺ and to 7.10^-3 for BaFBr:Eu²⁺.

It is clear that the CsBr:Eu²⁺ phosphor has a much better erasability than the commercial BaFBr:Eu²⁺ phosphor.

Read-out depth
Figure 2 shows read-out depth as a function of stimulation energy.

Fig 2: Read-out depth as function of stimulation energy per unit area for CsBr:Eu2+ (O) and BaFBr:Eu2+ ().

Much less light energy is needed to set the stored energy free as photostimulated emission in CsBr:Eu²⁺ than in BaFBr:Eu²⁺. Hence, less powerful light sources can be used in scanners for CsBr:Eu²⁺ and read-out depth can be higher, meaning that a higher gain CR system can be made.

4.2. Image quality
MTF
The MTF curves, shown in Fig. 3, demonstrate that the 150m CsBr needle IP prototype had a slightly lower sharpness than NDT film exposed with Se75 radiation, and slightly higher than NDT film exposed with Ir192 radiation. A commercially available powder IP (Agfa MD10) has a resolution that is comparable with NDT film exposed with Co60 radiation. The 500m CsBr needle IP leads to the same resolution as the MD10 powder IP at much higher absorption.

Fig 3: MTF for CR with a CsBr:Eu2+ needle IP and with conventional BaFBr:Eu22+ powder IP (MD10) in comparison with film MTF for several radiation energies.

Signal to noise ratio (SNR)
Using the mentioned exposure conditions and procedure an SNR value of 217:1 (23.4dB) was obtained for MD10. This value is not corrected for the Noise Equivalent Aperture (see further).

5. SNR comparison with Film and with an a-Se flat-panel system

For the CR system with the commercial MD10 plate and for the system with the NIP, the SNR was calculated using the cascaded system analysis (4). Input parameters were the detector absorption, the gain of the system, the Poisson excess noise and noise sources as X-ray quantum noise, conversion noise, screen structure noise and electronic noise.

The calculated SNR for the CR system with MD10 corresponded well with the measured value.

For film the SNR may be expressed as the ratio of film gain (G) to granularity (s_D) measured at the working density (density 2 above fog).According to standard EN584-1 the granularity is measured with a circular microdensitometer scanning aperture of 100 m.

For digital systems the SNR is defined as the ratio of the digital read-out intensity value I_meas to the standard deviation (s_D).

Normally the aperture to evaluate s_D is assumed to be equal to the pixel size of the digital system. However, for most digital systems, the actual aperture is not determined by the pixel size but by the light spread in the storage phosphor or scintillation layer. Therefore the SNR values should be normalised in order to compare with film systems.

The amount of light spread is reflected by the screen MTF. For film the 20% MTF value is reached at 10 lp/mm at standard NDT radiation conditions: 220kV–8mm Cu filter. The aperture that corresponds to a spatial frequency of 10 lp/mm is 100 m.

Therefore, we may define the noise equivalent aperture (NEA) as the aperture (expressed in m ) that corresponds to the 20% MTF value; e.g. if the 20% MTF value of a digital system is reached at 2,5 lp/mm, the NEA of this system is 400 m. Since we are concerned with the light collecting features of the individual pixels and not with the imaging characteristics of the system, the spatial frequency that determines the NEA may very well be situated beyond the Nyquist frequency of the system.

To compare systems with different NEA, the SNR values are normalised assuming Poisson statistics. For Poisson statistics we know that s_D is proportional to ÖN (N being the number of impacting photons). If a homogeneous and constant photon flux is provided, then N is proportional to the aperture area A = NEA². Hence Poisson statistics implies that s_D is proportional to NEA.

Next, we also correct for the square aperture used with digital systems versus the circular aperture used for Film.

We have compared the NIP based CR systems with the film systems shown in table 2 and with the a-Se flat-panel system (DRC). For this last system, the values have been taken from the literature (5). The results are shown in table 1a & 1b.

Detector	converter	Coating weight (mg/cm2)	a (%)	pixel size (m)	Fill factor	MTF (%)	Nyquist	MTF Nyquist (%)
DRC	a-Se	235	4.75	139	86	6.5	3.6	68
CR	NIP	50	3.53	140	100	3.5	3	30
CR	NIP	200	13.5	140	100	2.5	3	15
CR	MD-10	85	6.42	140	100	2.5	3	15
Table 1a: Digital detector characteristics.

Detector	Converter	SNR	MTF 20% (lp/mm)	NEA (m)	SNRcorr
DRC	a-Se	309	6.5	278	99
CR	NIP, 50m	259	3.8	286	80
CR	NIP,500m	328	2.5	370	79
CR	MD-10	217	2.5	400	48
Table 1b: Digital detector SNR.

Film	Dose D=2 (mGy)	g D=2	sD D=2
D2	29.0	5.4	0.015
D3	14.0	4.5	0.016
D4	8.7	4.4	0.019
D5	4.6	4.4	0.026
D7	3.2	4.4	0.031
D8	2.5	4.0	0.035
Table 2: Film Data.

The a-Se flat-panel system SNR value is for 20% of D7 dose exposure (ca. 74 mR on the detector), the CR system SNR values are for 50% of D7 dose exposure (ca. 185 mR on the detector).

The SNR results are depicted graphically in the Figure 4.

Fig 4: normalised SNR at 220 kVp + 8 mm Cu.

6. Conclusions

CsBr:Eu²⁺ is an excellent storage phosphor. It has better phosphor characteristics than BaFBr_1-xI_x:Eu²⁺ and it has the advantage that it can be grown in needles.

The emission and stimulation spectra of CsBr:Eu²⁺ are well suited for use in a CR system. CsBr:Eu²⁺ has a higher conversion efficiency and a lower stimulation energy, which makes deeper read-out possible at equal laser power.

CR performance can be improved significantly by using a needle screen instead of a powder screen. A needle screen yields images with acceptable resolution at higher coating weight and, which is more important for NDT applications to higher resolution at equal X-ray absorption. A needle screen is less turbid than a powder screen, which leads to lower values for the noise excess factor.

Use of a CsBr:Eu²⁺ needle screen in an existing CR system could lead to significantly improved image quality. Calculations demonstrate that SNR of a CR system with CsBr:Eu²⁺ NIP is higher than SNR of a CR system with BaFBr:Eu²⁺ powder imaging plates. The SNR is still slightly inferior to the SNR obtained with an a-Se flat panel system and is slightly better than the SNR obtained with D5 film for an exposure that is only 50% of the film exposure.

Acknowledgements

We would like to thank Luc Struye for the measurements.

References

1. M. Sonoda, M. Takano, J. Miyahara, H. Kato, "Computed radiography utilizing scanning laser stimulated luminescence", Radiology 148, pp. 833-838, 1983
2. W. Hillen, W. Eckenbach, P. Quadflieg, T. Zaengel, "Signal-to-noise performance in Cesium Iodide X-ray fluorescent screens", Proc. of SPIE Vol. 1443, Medical Imaging V: Image Physics, ed. R.H. Schneider, pp. 120-131, 1991
3. M. Thoms, "Photostimulated luminescence: a tool for the determination of optical properties of defect centers", Journal of Luminescence, 60-61, pp. 585-587, 1994
4. J.H. Siewerdsen, "Signal, noise, and detective quantum efficiency of a-Si:H flat-panel imagers", Dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy (Physics) in the University of Michigan, 1998
5. B. Rodricks, P. Soltani, B. Vaessen, P. Willems, "Optimization of a Se+TFT-based digital imaging system for wide-dynamic range NDT applications", to be published

© NDT.net - info@ndt.net

|Top|