The Effect of Direct X-ray on CMOS APS imager for Industrial Application

Transcription

1 The Effect of irect X-ray on CMOS APS imager for Industrial Application Kwang Hyun Kim a, b, Gyuseong Cho a a epartment of Nuclear Quantum Engineering, Korea Advanced Institute of Science and Technology, Kusong-dong, Yusong-gu, Taejon Korea b Hyun ae Nuclear Co., Ltd Bongchun-dong, Kwanak-gu, Seoul, Korea Abstract In this paper, we presented the effect of direct X- ray after scintillator on the CMOS APS imager using modulation transfer function (, noise power spectrum (NPS, and detective quantum efficiency (QE. 50 kvp of X-ray tube voltage at the SI of 300 mm were set with continuous mode micro focus X-ray machine on the assumption for industrial application such as PCB inspection. Lanex screen coupled CMOS APS imager was irradiated for long-term. From the experimental results, and also QE were degraded exponentially because of reduction of dynamic range caused by dark current or dark signal increase. For a given scintillator and an exposure condition, the degradation of image performance can be expected in case that the CMOS APS be used as a basic array. B I. INTROUCTION ecause of advantages of CMOS Active Pixel Sensor (APS over Charge Coupled evice (CC, [1] the CMOS APS imager is widely used in digital X-ray imaging as well as in general vision as an alternative instead of the CC. The CMOS APS generally consists of photo pixel array with an active amplifier within each pixel, row and column select logic, analog signal processor (ASP, AC, and timing and control logic block []. In a digital X-ray imaging, scintillator coupled CMOS APS imager (SC CMOS APS is used as one of a modality in radiography. The role of the scintillator is to convert incident X-rays into visible photon and its properties influence on the image performance as the CMOS APS itself does. Therefore, the optimization of both the scintillator and the should be performed to yield the best performance. However, since the optimization is fixed to the initial setup, the problem of long-term radiation exposure dependent can be overlooked. The condition of the long-term radiation exposure can be generated in non-destructive test (NT application. In that application, relatively high exposure mode X-ray machine is used, compared to medical X-ray imaging. This condition results in lots of exposure to the image array itself by the transmitted X-ray through scintillator (direct interaction with the since it gives high flux of radiation photons to obtain the clear image of rather thick objects. In the case of the scintillator coupled CMOS APS under the X-ray irradiation, the amount of absorbed dose or dose rate in CMOS APS imager is more or less different than Co-60 direct irradiation on the imager, and the result of that can be different correspondingly. Even though it is known that the CMOS APS has more radiation tolerance than the CC, the CMOS APS also is not free from radiation environment. Generally, Co-60 irradiation has been used to evaluate radiation response of the CMOS APS. Through the ionization mechanism in the, overall radiation effects on the result in increase leakage current by increasing radiation dose before system failure [3]. In this paper, we mainly present the image performance influenced by the direct X-ray for the SC CMOS APS imager. By using Lanex screen scintillator and CMOS APS imager, we investigate changes of image quality factors of the detector such as modulation transfer function (, noise power spectrum (NPS, and detective quantum efficiency (QE under the X-ray condition for the NT application. II. LONG TERM EXPOSURE ON SC CMOS APS IMAGER A. Experimental Set-up To see the transmitted X-ray effects on the detector, we used two scintillators of Lanex TM Fine and Regular and two CMOS APS array of RadEye TM. The RadEye CMOS APS has an active area of 5mm by 50 mm with the pixel size of 48 µm. Fein focus TM X-ray machine with Tungsten target and Beryllium window was used in our tests. Cumulative exposure was measured using an ionization chamber of RA CHEK TM PLUS (models The properties of Lanex screen and the imager were explained in Table 1. Manuscript received October 9, 003. Kwang Hyun Kim and Gyu Seong Cho are with the KAIST Kusongdong, Yusong-gu, Taejon , Korea (telephone: , e- mail: /04/$0 004 IEEE. 145 Authorized licensed use limited to: IEEE Xplore. ownloaded on October 1, 008 at 04:0 from IEEE Xplore. Restrictions apply.

2 Table1. Scintillator coupled CMOS APS and its test conditions. X-ray Exposure Conditions Tube Voltage : 50kVp Tube Current : 500uamp Thickness: 34 & 67 Scintillator ensity: 7.34 Properties (g/cm 3 for Fine & Regular (mg/cm CMOS APS Imager Specification Active Area:5mm x 50 mm Pixel Size: 48um Noise floor: 150 electrons rms igitization noise: 50~300 electros rms B. irect detection, SNR, ynamic Range Except irradiation condition of the X-ray, the direct detection in the scintillator coupled detector is highly dependent on the properties of the scintillator itself. Concerning the physical properties of the scintillator, we can simply summarize into two characteristics of high light output from the scintillator and good resolution. These two characteristics are always conflicting in the design of the X-ray imaging system. The thicker scintillator has the more light output but the worse resolution since it has blurring effect by the light scattering in the scintillator itself. One of solutions to overcome this conflict is a tradeoff between the light output and the resolution. From that tradeoff process, the thickness that absorbs 85% of an incident radiation is accepted as an optimum thickness as a rule of thumb. However, this design rule leads direct X-ray interaction on the and has a chance to change the properties of the and overall image quality since the rest of 15%, the transmitted X-ray, interacts with the directly. To estimate the transmitted X-ray of the scintillators, we did Monte Carlo simulation by using semi-empirical X-ray spectrum [4] at 50 kvp tube voltage. The amount of the transmitted X-ray under that condition was 43 %, 8 %, and 16% for Lanex fine, Lanex regular, and Lanex fast, respectively in a singe X-ray shot. The increase in quantity is highly dependent on the exposure time and exposure rate. Related to the direct in the detection scintillator coupled detector, the analyses of CC based detector have been published by others using experimental measurements and semi-empirical analysis [5]. However, since these analyses have focused on only short-term single exposure on the detector under the medical condition, those approach and their research results cannot satisfy CMOS based detector under long-term exposure. In a single short-term exposure of the CMOS based detector, a pixel signal is made not only from the scintillation light on to the photo but also due to the direct interaction of X-rays, so S S tot light X, (1 where the S light and the S X are the signals, the scintillation light and the direct X-rays. The signal to noise ratio (SNR can be calculated from the acquired image by S / N S tot/ σ ( S tot S and the σ ( tot S where the tot are the mean image signal and the root mean square of the total signal. And, a workable definition of dynamic range (R is the ratio of the highest signal which a detector can record to the lowest (darkest signal as follows. Using equation ( and (3, we calculated the SNR and the R and plotted the results as shown in figure 1. In the calculation they were all normalized with an initial value of kr to show relative increase and decrease of the SNR and the R for exposure. In this figure, although there is an increase of the SNR in contradiction to the R, it cannot be accepted as a proper explanation since the signal component contains dark signal generated by direct X-ray detection. Normalized value (arb. unit R Log ( S highest / S 10 darkest R SNR 5 6 Entrance Cumulative Exposure (kr Fig.1 Normalized SNR and R based on measured SNR and R in AC unit at each cumulative exposure level. So as to see the extent of dark signal we plotted total signal, dark signal, and net signal (total signal minus net signal as shown in figure. These signals were obtained by a Lanex fine coupled the at on integration time of 1100 msec under the condition of 50 kvp and 500 µamps at each cumulative exposure and they were normalized also with an initial value of kr. The dark signal increased by each cumulative exposure and contributed directly the total signal. Therefore, we can conclude that the dark signal influences on the increase of the SNR as an additional signal component but not real and the decrease of the R as base line noise. ( ( /04/$0 004 IEEE. 146 Authorized licensed use limited to: IEEE Xplore. ownloaded on October 1, 008 at 04:0 from IEEE Xplore. Restrictions apply.

3 Normalized Signals (arb. unit Entrance Cumulative Exposure (kr Fig. Normalized total, net, and dark signal based on measured values in AC unit at each cumulative exposure level. Total Signal Net Signal ark Signal C. Modulation Transfer Function ( The modulation transfer function ( is a convenient factor to assess the image resolution and is expressed in terms of the resolving power of the imaging system [6]. For the scintillator coupled array, system considering both resolutions of the scintillator and the should be evaluated since the scintillator in front of the CMOS array influences on the resolution in the system. sys sc int where the sc intand the are the scintillator and the, respectively. The calculated and the measured system at an initial of kr are shown in figure (3. In the calculation of system, Lanex Fine intrinsic [7] and 48 um pixel size with 85% fill factor. From the acquired line spread-function (LSF by using a Tantalum phantom, 1.5 mm of thickness, with a 10 µm width slit, the was calculated by Fourier transform. (4 The discrepancy between calculated and measured is based on the fitting problem of intrinsic Lanex Fine which was adopted from other literature. However, they are comparatively agreed well. Now, we again measured each corresponding to each cumulative exposure and showed the results in figure Spatial Ferequency (mm -1 kr kr.0kr Fig.4 Measured variation for each cumulative exposure level. The system was degraded from initial value by each cumulative exposure. From the figure 4, we plotted degradation of normalized at an initial value of kr by using following equation. deg where f Nyq. is Nyquist frequency and R is exposure level. The degradation of the followed exponential decay by increasing exposure level as shown in figure 5. f Nyq 0 f Nyq 0 ( f df ( f df _ R var. _ R 0., Model: ExpLinear y p1*exp(-x/p + p3 + p4*x (5 Measuerd Calculated Normalized Spatial Frequency (mm -1 Fig.3 Comparison between calculated and measured at an initial exposure. Entrance Cumulative Exposure (kr Fig.5 egradation of from an initial for each cumulative exposure level. These data were calculated by equation (5 The at kr was steeply degraded from the initial value. Above kr, the degradation of the is rather /04/$0 004 IEEE. 147 Authorized licensed use limited to: IEEE Xplore. ownloaded on October 1, 008 at 04:0 from IEEE Xplore. Restrictions apply.

4 gradual at each higher exposure level. This figure decisively shows that the detector resolution is dependent on the amount of the direct X-rays on the detector as well as intrinsic resolution of scintillator and.. Noise Power Spectrum (NPS The noise transfer properties of the imaging system are given by noise power spectrum (NPS with the spatial frequency domain. NPS measurement were performed for the imager employing each scintillator up to ~ 50 % of pixel saturation. In this measurement all data were acquired at the SI of 300 mm. For the flood field data offset and flat field correction were performed by others recommended method [8]. In order to detect and eliminate the defective pixel in the image region, a 3 x 3 median filter was used, resulting in less than 1 % for over all pixels. Finally, NPS was computed by Fourier transform of 18 x 18 sub image, which were region of interest (ROI from the central region of the images, and the results were normalized by [ (, ] FT flat field x y x y (6 NPS ( u, v, ( m of 18 x 18 ROI N N x y FT flat field(x, y represents the ensemble where [ ] average of the squares of the magnitude of the Fourier transformed 18 x 18 sub image, N x and N are the y number of pixels in the x and y directions, respectively, and and x are the pixel pitch of 48 µm in x and y directions, y respectively. m is mean signal of the 18 x 18 sub image. And then from equation (6, 1 NNPS was estimated using a radialaveraging technique. Normalized NPS (mm 4.0x x x10-6.5x10-6.0x x10-6 x Spatial frequency (mm -1 kr kr.0kr Fig.6 Measured NNPS at each cumulative exposure level showing slightly increase of NNPS at high frequency. From the acquired images corresponding to each cumulative exposure, small reduction of the NPS was revealed at low frequency, but the extent of the NPS change was not severe in the long-term exposure effect as shown in figure 6. E. etective Quantum Efficiency (QE The detective quantum efficiency (QE describes the transfer of the signal to noise ratio, SNR, through the imaging chain and is regarded as the most useful measure of sensitivity and noise performance of an imaging system. As a function of the spatial frequency of the object details, the QE (f is defined as QE( f ( f where q is a calculated average photon number incident onto the detector area by using 50 kvp spectrum from [4] and the photon fluence per exposure for the entire spectrum [9]. Figure 7 shows the calculated QE (f at each exposure. As the cumulative exposure increases, the QE (f increases at the low frequency region, but at the high frequency region the QE drops. QE Spatial Frequency (mm -1 kr kr.0kr Fig.7 Calculated QE at each cumulative exposure level using measured, NPS, and photon fluence. We plotted degradation of the QE at an initial value of kr by using same relation as in equation (5, and showed the results in figure (8. The degradation of the followed exponential decay by increasing exposure. Normalized Value (arb. unit q NNPS( f, QE Exponential ecay Fit Entrance Cumulative Exposure (kr Fig.8 egradation of QE from an initial for each cumulative exposure level. These data were calculated by same relation as in equation (5. ( /04/$0 004 IEEE. 148 Authorized licensed use limited to: IEEE Xplore. ownloaded on October 1, 008 at 04:0 from IEEE Xplore. Restrictions apply.

5 III. ANALYSES AN ISCUSSION The direct X-rays that are transmitted through scintillator cause the dark signal increase in the detector by cumulative exposure. The increased dark signal again degrades the dynamic range and the resolution of the detector. The initial system follows the resolving power of the as follows: where So max and So min are signals with open aperture and opaque aperture, respectively. However, increasing irradiation for long-term exposure leads to increase the dark signal ( S as a base signal source. Therefore, the S can be treated as a separate term being added to the So min to the S o max as given in the follow equation. or (8 (9 (10 Since there is a correlation between the and the QE as expressed in equation (7, eventually, the QE degradation follows the trend for a fixed NNPS. Figure (9 shows each correlation as a function of exposure level. Final form of QE related to the exposure can be expressed as follows. (11 where a is constant and R depends on the extent of the transmitted X-rays through a given scintillator properties such as attenuation coefficient and thickness. Normalized value (arb. unit S S ( S ( S S omin omin ( S + ( S o min o min S Somin S + ( So max omin QE( R QE R 0 e ar ark Signal ynamic Range QE IV. CONCLUSION AN FURTHER STUY We showed and explained the overlooked cumulative exposure effect in a scintillator coupled CMOS APS for X-ray imaging by measuring, NPS, and QE. and also QE were followed by the cumulative exposure, which were transmitted through scintillator, and mainly were degraded exponentially. This was because of reduction of dynamic range caused by dark current or dark signal increase. For a long-term and an improved design of the detector for the NT application, careful analysis of image evaluation parameters should be considered. For a given scintillator and an exposure condition, the degradation of image performance can be expected in case that the CMOS APS be used as a basic array. Experiments matrix such as different scintillator and the higher exposure dose will be performed to complete the effect of radiation on the image performance in the future. ACKNOWLEGMENT This work was supported in part by the Ministry of Science and Technology of Korean Government under National Midand Long-term Atomic Energy R& Program in 003 V. REFERENCES [1] E.R. Fossum, CMOS IMAGE Sensors: Electronic Camera-In-A-Chip, IEEE Trans. Elec. evices, vol. 44, No. 10, pp , Oct [] E.R. Fossum, CMOS active pixel image s, Nucl. Instr. and Meth. A 395, pp91-97, [3] G.R. Hopkinson, Radiation Effects in a CMOS Active Pixel Sensor, IEEE Trans. NS. vol. 47, No., 6 pp [4] Peter Hammersberg et al, Absolute energy spectra for an industrial micro focal X-ray source under working conditions measured with a Compton scattering spectrometer-full spectra data, Linkoping Electronic Articles in Mechanical Engineering Vol. 1 (1998. [5]. G. arambara, Image-quality performance of an a-si:h based X-ray imaging system for digital mammography, Nucl. Instr. Meas. A 477 (00 pp [6] K. Rossmann, Ponit Spread-Function, Line Spread-Function, and Modulation Transfer Function, Radiology 93, 57-7 (1969. [7] J. H. Siewerdsen, Signal, Noise, And etective Quantum Efficiency of a-si:h Flat-Panel Imagers, in Ph thesis of the University of Michgan Chap.5. pp. 119, [8] Srinivasan Vedantham et al, Full breast digital mammography with an amorphous silicon-based flat panel detector: Physical characteristics of a clinical prototype, Med. Phys. 7 (3, March 000, pp. 558~567. [9] J. M. Boone, X-ray Production, Interaction, and etection in iagnostic Imaging Handbook of Medical Imaging (SPIE, 000, edited by SPIE, Vol. I, Chap. 1, pp. 50. Entrance Cumulative Exposure (kr Fig.9 Relation among dark signal, dynamic range,, and QE as a function of cumulative exposure level /04/$0 004 IEEE. 149 Authorized licensed use limited to: IEEE Xplore. ownloaded on October 1, 008 at 04:0 from IEEE Xplore. Restrictions apply.