Optimization of scintillator thickness for Dual X-ray imaging system 1,2 Han Gyul Song and 1,2 Kwang Hyun Kim * 1 Biomedical Engineering, Jungwon University, Republic of Korea 2 Basic Atomic Energy Research Institute (BAERI), Jungwon University, Republic of Korea II. Materials and Methods IV. Conclusion & Further Study III. Results & Analysis In this study, the scintillator we chose was CsI(Tl). Its thicknesses are 0.16, 0.23, 0.3, 0.4cm and DRZ phosphor screens, Std, Plus, High. The intensity of the X-ray energy is set by a 100kVp energy spectrum and 1, 2, 3, 4mA with 65cm SOD (Source to Object Distance), which is generally used for bone density measurements. The filters for separating low and high energy are Er 0.5mm, for the low energy X-ray and, Cu 0.8mm with Rh 0.4mm attachments for the high energy X-ray. Then, to measure the X-ray dose, we used an ion chamber on both sides of the collimator and on the detector. By using this method, we checked and compared the exposure dose at both the non-filter state and the filter applied state. The ion chamber we used for the experiment is the Dosimeter 2186 made by Radcal Inc. The experiment process consisted of two broad steps. At first, we examined the exposure dose without the filters and the collimator. Next, we applied the filters and the collimator and conducted the procedure again. From this, we could compare exposure doses between the non-filter procedure and the applied filter procedure. To check the ADC transmission values, we used the Radeye1 EV Image Sensor Premium Grade made by Rad-icon Inc, which is a CMOS sensor. Because this sensor is vulnerable to static electricity, we wore grounded gloves and bands during the experiment. To block external light, we turned off all the lamps in the room and worked only at night time. I. Introduction  A dual X-ray imaging system uses high and low energy X-rays at the same time. It is applied to check body fat, for bone density diagnosis, and growth plate prognosis on account of the advantages of this system. First it creates a clearer image and it also reduces the X-ray’s hardening effect.  Currently, the daul X-ray imaging system is materialized by two methods. One uses two X-ray sources from the beginning whereas; the other uses a moving filter. However, both methods have some problems. The former, has a relatively bulky volume and the production cost is high, therefore it is hard to commercialize. The latter, has uncertainty which is caused by the moving filter. In addition, existing equipment makes irregular output under the same examination conditions in both methods.[1, 2, 3]  In this work, we suggest a new dual X-ray imaging method to increase the precision of the equipment. We applied filters on a collimator to separate two X-ray energies, low and high, from a single energy X-ray beam. We conducted an experiment that examined the X-ray exposure dose and the ADC transmission values to investigate the performance of the filters. Then, according to inductions from a previous study, we optimized the thickness of the scintillator for low and high energy. Figure 1 shows the result of spectrum simulation for choosing the filter. Figure 2 demonstrates the R value of the filter combinations. Figure 1 and 2 induced in previous study[4]. It shows Er 0.5mm and Cu 0.8mm + Rh 0.4mm has minimum R value. Therefore, we selected this filters. Figure 3 demonstrates the performance of the Er filter. Because the X-ray was entered in the form of a spectrum, this figure means 1~60keV low X- ray energy even if 100keV energy was entered. Figure 4 demonstrates the performance of the Rh and Cu filters. These two graphs demonstrate the same tendency as with the previous study that simulated tendency. The tendency is that the low energy gotten by the Er filter has a higher ADC value and the high energy gotten by the Cu and Rh filter has a lower ADC value. Figure 5 shows the exposure dose that was applied to the filters at 100kVp. It demonstrates that the exposure dose highly decreased after we applied the filters. Before we applied the filters, the exposure dose was 443.7mR at the X-ray hole, and 69.91mR at the top of the detector. From this result, we got a high performance from the filters for reducing the exposure dose. Figure 6 demonstrates the optimum thickness of the CsI(Tl) at low energy. We got 0.16cm CsI(Tl) as the optimum thickness for low energy. It is the same result as with our previous simulation result. Figure 7 demonstrates the optimum thickness of the CsI(Tl) at high energy. We got 0.23cm CsI(Tl) as the optimum thickness for high energy. It is, again, the same result as with the previous simulation result. Throughout this study, we investigated the performance of the filters and the optimum thickness of the scintillators for low and high X-ray energy used by a CMOS sensor and an ion chamber. After we applied the filter, the exposure dose significantly decreased. The exposure dose was 1/100 of the level compared to what it was before we applied the filter and the X-ray hardening effect decreased, too. Then, we got the result that 0.16cm CsI(Tl) is the optimum thickness for low energy and 0.23cm CsI(Tl) is the optimum thickness for high energy. Therefore, we investigated our previous study and filter performance. We will apply this result to actual dual X-ray imaging detectors and imaging systems. Figure 3. Low energy that applied Er filter DRZ screen ADC value at 100kVp Figure 4. High energy that applied Ru, Cu filter DRZ screen ADC value at 100kVp V. Reference 1. Reid DM, Lanham SA, McDonald AG. Osteoporosis, pp ;2. 2. Gundry CR, Miller CW, Ramos E, Moscona A. Stein JA, Mazess RBOsteoporosis, Radiology, pp , Genant HK, Jour. of Bone Miner. Res., pp ;9. 4. Kwang Hyun Kim, X-ray Filter Design and Its Evaluation in Dual x-ray Absorptiometry(DXA), Vol 57, No.4, pp , Figure 6. Low energy that applied Er filter CsI(Tl) scintillator ADC value at 100kVp Figure 7. High energy that applied Rh, Cu filter CsI(Tl) scintillator ADC value at 100kVp Figure 5.Exposure dose at 100kVp when filter is applied Figure 1. Simulated spectrum when filter applied Figure 2. Simulation result of the R value for the suggested K-edge method