Validation of virtual spectrometer created in RADlab1.03.

Spectrometer is used to perform various in vitro tests. The ability to successfully perform these tests depends on technologist's skill. Therefore, proper training of technologists is mandatory in gamma spectrometry. During the training, they need to have sufficient practice to gain sound theoretical and practical knowledge. High cost of spectrometer and risk of their damage during independent practice may hamper the process of proper training. Hence, there is a need of cheaper and more feasible option. Virtual spectrometer created in RADlab1.03 can address this issue. The immediate objective of this paper is to validate this virtual spectrometer so as to be used as an educational and research tool for trainees. Virtual spectrometer was calibrated using Cs-137 standard source and Cs-137 spectrum was recorded by positioning 28106 Bq Cs-137 source at 2.35 cm above top surface of the well, 1 cm above from the bottom of the well and at the bottom of the well. Ba-133 and Co-60 spectrum were also recorded. The experiments were repeated with real spectrometer for exactly the same conditions as applied to the virtual spectrometer. The paired t-test was applied to find the difference in mean photopeak at 5% level of significance. The sample data provided satisfactory evidence that mean photopeak obtained with real as well as virtual spectrometer were same at P value of 4.641 × 10(-4),1.57 × 10(-12),1.40 × 10(-24), 1.26 × 10(-16), and 8.7 × 10(-9) for Cs-137 (photopeak: 664 keV, Co-60 (photopeak: 1181 keV), Co-60 (photopeak: Co-1348 keV), Ba-133 (photopeak: 304 keV) and Ba-133 (photopeak: 364 keV) respectively.

INTRODUCTION

The spectrometer is the equipment used in variety of quantitative tests such as assessment of Glomerular filtration rate, radio immune assay, and life span of red blood cell. These tests are simple and yield highly accurate results in relatively less time with good performance of equipment. However, the accuracy of the results depends on operator's skill.

Automation of spectrometer has improved the performance, and now with just a push button more reliable results can be obtained in less time. However, the automation has increased the cost of spectrometer and made it less flexible from the user end.

Students/trainees of-nuclear medicine must be familiar with significance and limitations of the in vitro tests. They need to be familiar with the factors that influence the counting efficiency. In general, trainees are in groups of ten to twenty and at least one spectrometer per two trainees is required in practical class room so that they can follow the instructions and do the practical simultaneously. This will infuse more confidence in them. This will also give them more chance to interact with the spectrometry system independently. However, the high cost of the spectrometer and associated accessories for practical demonstration does not make this option feasible and besides this there is a risk of damage of equipment during practice. Moreover, automation does not allow many variables (for example effect of shielding material around the detector) to change and demonstrate its effect on the result. There is a need to search for a cheaper solution of spectrometer where trainees performing experiments in the practical class room and also practice independently without risk of its damage. Virtual spectrometer holds promise as the software can be loaded on personal computer/laptop, which are more commonly available and is cheaper option compared with real spectrometer. Dagistan sahin developed virtual simulation lab called RADlab1.03, based on Monte Carlo simulation.[1] It is easy to create virtual spectrometer with the help of graphical user interface and their study claims that variety of experiments required to achieve the education objectives can be demonstrated with it. Simulation of detector response function is a difficult task,[23] and simulating spectrometer will be more complex since it involves simulation of many pieces of electronics such as high voltage supply, radiation sources, preamplifier, amplifier and multi-channel analyzer (MCA) and their effects on spectrum. The reference available with the RADlab1.03 documentation does not include validation of the developed tool and especially it is documented that the tool has not been developed by the professional team of software developers. This study validates the virtual spectrometer created in RADlab1.03.

MATERIALS AND METHODS

RADlab1.03 is a virtual simulation laboratory developed for radiation detection and measurements and is free to download and use. Radiation sources, radiation detectors, electronics components, and counting accessories modeled using Monte Carlo simulation techniques are available as graphical objects. User can select and wire them properly for a simulation of a particular experiment. The interaction of radiation with detectors have been modeled using analog Monte Carlo engine in Java™ runtime environment.

Java™ runtime environment was installed on a laptop having window 7, 32-bit operating system, Intel(R) Pentium(R) CPU B960 @2.20 GHz processor and 2.00 GB (1.85 GB usable) RAM. Then RADlab1.03 was installed. The Biodex Atomlab950 spectrometer was modeled in RADlab1.03. The spectrometer consist of 2” × 2” well type cylindrical NaI (Tl) crystal detector, high voltage supply, preamplifier, amplifier and MCA.

In order to calibrate the virtual spectrometer, Cs-137 (28106 Bq) source was placed at 0.5 cm from the surface of NaI(Tl) detector. A MCA was set up for 2048 channels and 60 s real time to acquire the spectrum. The amplifier setting was kept at the minimum value (coarse gain = 10, and fine gain = 10). The voltage applied to the detector was switched on and then gradually increased while awaiting spectrum to appear and then disappear on MCA. The high voltage and the corresponding integrated count of the spectrum were recorded from appearance of the spectrum to its disappearance; the high voltage corresponding to the maximum count was selected as operating voltage of virtual spectrometer. Then, keeping voltage constant, amplifier coarse gain was adjusted to move photo peak near channel no 662 (662 keV) and further fine gain was also adjusted so that photo peak appeared on the channel number 662.

After calibration, 25 times background subtracted Cs-137 spectrum (N = 25) were recorded by positioning 28106 Bq Cs-137 source 2.35 cm from the bottom of the well, and analyzed for energy peaks. Similarly, 25 times background subtracted Co-60 and Ba-133 spectrum were also recorded. Mean and standard deviation of each energy peaks were calculated. The spectrum was also recorded by positioning the Cs-137 at two different positions one at 1 cm from the bottom of the well and another at the bottom of the well.

Besides the above, virtual spectrometer was also verified for the photopeak shift with amplifier gain, appearance of detector output on oscilloscope, facility for integrated ROI counts, variation of photopeak counts with activity or with distance from the source/with geometry of the source.

The initial activity of Cs-137, Co-60 and Ba-133 source available in the department was not traceable. The unknown activity of Cs-137 was determined with a known activity of another Cs-137 source (activity 0.1041 μCi (3.852 kBq) on ref date: 1 feb 2011 from Eckert and Ziegler, Isotope products, California). In order to validate the virtual spectrometer, 25 times background subtracted the Cs-137, Co-60, and Ba-133 spectrum were also acquired with Atomlab 950 Medical spectrometer (Bio-dex Medical systems, Brookhaven R and D Plaza, New York, Model No. 187–950) for exactly the same conditions as applied to the virtual spectrometer. The real spectrometer has 2 inch × 2 inch cylindrical NaI (Tl) crystal detector and the diameter and depth of the well are 0.75 inch and 1.44 inch respectively. The cover (lid) of the well is made of 0.125 inch thick Pb (lead). Detector is shielded with one inch thick lead from all side to minimize the contribution of background radiation. The MCA has 1024 channel. This spectrometer has automatic calibration facility, it adjusts the high voltage and amplifier gain so that Cs-137 peak falls between the channels number 400–500. It was calibrated by positioning Cs-137 disc source on the top of the well counter following manufacturers instruction. After the calibration, 28106 BqCs-137 (type R2300A SRL 349) gamma reference standard source from Electronic Corporation of India Ltd. was used for recording and analyzing the spectrum. For comparing the error between the theoretical peak, virtual and real peak, the theoretical value of Compton Edge Peak and backscatter peak was calculated using the formula (1) and (2) respectively.[4]

Where E is in keV.

We applied paired t-test on the sampled data from real and virtual spectrometer and set up the null hypothesis at 5% level of significance that there is no difference between the mean photopeak position obtained with virtual and real spectrometer and the observed difference are random distribution from photopeak obtained with a real spectrometer with zero mean. Excel 2010 (Microsoft Corporation, USA) was used for paired t-test analysis.

RESULTS

Figure 1b shows how various pieces of equipment were wired to create virtual Spectrometer in RADlab1.03. The spectrometer consist of high voltage supply, NaI(Tl) well counter, preamplifier, amplifier, and MCA. High voltage supply unit has one output port to transmit high voltage, NaI(Tl) detector has one input port to receive high voltage and one output port to transmit the electric pulse formed as a result of conversion of light photon into electric by PMtube in the real spectrometer. Interaction of gamma radiation with detector material yields light photons, In RADlab1.03, detector and PMtube functionality are combined and supplied as one unit as NaI(Tl) detector unit. Preamplifier has two input port one to receive signal and another to receive test input signal and two output port one to transmit signal and another to connect with PreAmplifierVoltageSignalCable with amplifier, Amplifier has two input port to receive input signal and another to connect with PreAmplifierVoltageSignalCable and two output port to transmit bipolar and unipolar signal. MCA has one input port to receive the signal and display output as energy spectrum. It is also possible to obtain the energy spectrum using single channel analyzer, the experimental set up in this case is shown in Figure 1(d).

Figure 1

(a) spectrum as multi channel analyzer (MCA) output, cursor, left and right marker, the position and channel number energy, counts displayed at the bottom of the spectrum. (b) Experimental set up showing proper connectivity how detector, high voltage supply, preamplifier, amplifier and MCA are wired for the desired output. (c) Cs-137 spectrum plotted using Microsoft Excel. (d) Experimental set up showing how the detector module, high voltage supply, preamplifier, amplifier, single channel analyser and counter are wired to simulate a medical spectrometer

Operating high voltage was found to be 900 V for virtual spectrometer. No counts were observed until high voltage reached 850 V, thereafter counts increased gradually with increase in voltage and attained maximum at 900 V and decreased thereafter to zero at 940 V. The amplifier coarse and fine gain were found to be 30 and 90 respectively for all isotope (Co-60, Cs-137, and Ba-133). Whereas, the operating voltage in case of real spectrometer was found to be 1039 V and the amplifier gain was in the symbolic form such as Gain 1, Gain 2, Gain 4 for Co-60, Cs-137, and Ba-133 respectively.

In virtual spectrometer, with increase in amplifier gain, photo peak shifted towards the higher energy side of the spectrum [Table 1]. The appearance of photo peak, Compton edge, Compton plateau, Compton valley were observed in both virtual and real spectrum as shown in Figures 2 and 3. The peak at 31 keV were seen in both virtual [Figure 1c] and real spectrum [Figure 2b]. Few additional peak counts at 60 keV was observed in the simulated spectrum, and 71keV peak observed experimentally.

Table 1

Variation of photo peak with amplifier gain observed with simulated spectrometer in the Cs-137 spectrum

Figure 2

(a) Cs-137 virtual spectrum observed at 661 keV (0.0015% error from theoretical). (b) Cs-137 real spectrum. (c) Co-60 virtual spectrum photopeak observed at 1038 (7.17% error from theoretical), 1148 (2.13% error from theoretical) and 1303 keV (2.25% error from theoretical) respectively. (d) Co-60 real spectrum photopeak observed at 1027 (8.15% error from theoretical), 1173 (0.0% error from theoretical) and 1335 keV (0.15% error from theoretical) respectively

Figure 3

(a) Ba-133 virtual spectrum photopeak observed at 314 (11.80% error from theoretical) and 368 keV (4.12% error from theoretical) respectively. (b) Ba-133 real spectrum photopeak observed at 314 (11.80% error from theoretical) and 369.4 keV (3.76% error from theoretical) respectively. (c) Experimental set up source at the surface of the well. (d) Cs-137 spectrum obtained with the real spectrometer shown in Figure 3c

Five peaks at or around 5,30,60,427,661 keV were observed in virtual spectrum their mean and standard deviation are given in Table 2, whereas six peaks at or around 38, 78, 186, 217, 441, and 664 keV were observed in real spectrum their mean and standard deviation are also given in Table 2. The percentage error from theoretical backscatter, Compton peak, and photopeak, in real and virtual spectrum for Cs-137 is given Table 3.

Table 2

Mean energy peaks of Cs-137 and their SD observed in the virtual and real spectrum

Table 3

Percentage error difference from theoretical data for backscatter, Compton and photo peak in real and virtual spectrum for Cs-137

With real spectrometer, the graphic signature of the Cs-137 spectrum obtained with lead lid cover by positioning the source at the bottom and 1 cm above from the bottom are same. However, the energy resolution (5.6%) and photopeak [560–760 keV] counts (29688) were better when the source was at bottom in comparison to (6.3%, 26445) when the source was 1 cm above from the bottom. Without lead lid cover when Cs-137 source was positioned at the bottom FWHM and photopeak counts were 4.3% and 30098 respectively. It was also seen that when source was placed at the bottom, the Compton peak was only visible in simulated spectrum whereas backscatter was observed in real spectrum.

Several functions such as background measurement, energy calibration and spectrum recording and saving the spectrum data in ASCII format that can be imported in Excel is available with virtual spectrometer. Spectrum can be depicted graphically and evaluated via region of interest.

Virtual spectrometer has the facility to display counts and their corresponding position (in KeV) on the curve as the cursor are placed on it [Figure 1a]. It also has the facility to save the spectrum data that can be imported in Microsoft Excel for further analysis [Figure 1c]. The Atomlab 950 spectrometer do not provide export and save option.

Comparatively less count in the real spectrum was observed in comparison with virtual spectrum. Backscatter peak and downward slope in Compton continuum in real spectrometer was observed in the real spectrum, on the other hand backscatter peak was absent and a relatively flat Compton continuum was observed in virtual spectrum [Figure 4].

Figure 4

Real and virtual spectrum when the source was positioned at the bottom and 1 cm above from the bottom

The result of paired t-test for the difference of mean photopeak position obtained with real and virtual spectrometer as result of N = 25 measurements are given in Table 4. At 5% level of significance, the sample data provided satisfactory evidence that mean photopeak obtained with real as well as virtual spectrometer were same at P value of 4.641 × 10−4,1.57 × 10−12,1.40 × 10−24, 1.26 × 10−16, and 8.7 × 10−9 for Cs-137 (Photopeak: 664 keV, Co-60 (photopeak: 1181 keV), Co-60 (photopeak: Co-1348 keV), Ba-133 (photopeak: 304 keV) and Ba-133 (photopeak: 364 keV) respectively. The photopeak position mentioned in bracket in the previous sentence refers to the photopeak position obtained with real spectrometer.

Table 4

The value of paired t-test calculated from the sample obtained with real and virtual spectrometer for Cs-137, Co-60 and Ba-133 and t-critical at 5% level of significance

DISCUSSION

Virtual spectrometer can be used instead of real spectrometer for training purpose by installing RADlab1.03 on a laptop. Our results show that the virtual spectrometer is in agreement with real spectrometer. Therefore, virtual spectrometer can be made available for the students/trainees.

Output of real spectrometer is dependent on the high voltage supplied and gain of the amplifier. Our results show that output of virtual spectrometer is dependent on the applied high voltage as is the case with real spectrometer [Table 1] and is in agreement with the text book results.[5]

Virtual spectrometer was calibrated with standard sources of Cs-137, Ba-133 and Co-60. Spectrum have been obtained and Virtual spectrometer was found to be linear [Figures 2, Figure 3a and b]. It also has the facility to obtain photopeak count [Figure 3d]. Thus except the determination of critical volume it can be used for all practical demonstration recommended for nuclear medicine students/trainees by Early and Sodee.[6] The determination of critical volume requires the radioisotope to be diluted with water in graded fashion/volume therefore, cannot be demonstrated with virtual spectrometer.

Photo peak, Compton peak and Compton continuum are in observed with both virtual and real spectrum. However, downward slope of Compton continuum, backscatter peak, and lead X-ray peak in real spectrum found was due to lead surrounding the detector.[7] The detection of photons scattered by more than 90° from the shielding material increases the lower side of the Compton continuum. The detection of photons scattered at 180° from shielding material give rise to backscatter peak. The interaction of gamma rays with k-shell electron of lead produces fluorescent X-rays.[7] The detection of these fluorescent X-ray produces a small peak at 72 keV [Figure 2b].

We have observed Compton peak both when source was either inside the well or outside the well in real as well as virtual spectrometer experiment. However, it was not distinguished visually and likely masked by scattered photons when the source was inside the well [Figure 4a and b] when compared to when the source was outside the well in case of real spectrometer [Figure 2b]. The possible explanation for this observation could be due to the high geometric efficiency of the source and the source was surrounded by the one inch thick lead. Due to high geometric efficiency,[8] maximum number of emitted gamma photons from source deposits their energy in detection volume and those photons that escaped initially and re-entered in the detection volume leading to increase in counts. This phenomenon may cause coincidence loss[9] and finally reduction in photopeak counts was observed.

With real spectrometer, the decrease in number of counts found was due to entry of scattered photons from shielding material in the detection volume. This increases the number of photon per second to be detected and because of coincidence time loss less counts than the original number of photons are observed. There is no shielding material surrounding the detector in case of virtual spectrometer. And here also similarly the absence of lead surrounding the detector is the explanation for observance of no peak at 72 keV. The peaks at 59.12 keV may be the result of the summation of two peaks of 30 keV K-xray from the Ba-137 and another the Iodine peak at 30 keV.

We used only MCA to record the spectrum and not the SCA to reduce the volume of the data so that effectively it can be explained in one article. Further research should include validation of virtual spectrometer constructed with single channel analyzer.

The study did not include other widely used radiation source in nuclear medicine such as Tc-99 m, Lu-177, I-131, F-18, Ga-68 and their spectrum. These radiation sources are not available in the RADlab 1.03 and this is the limitation of the RADlab1.03 also from the nuclear medicine prospective, because students cannot practice the experiments involving these radiation sources.

This study has relevance to Instructor and trainees. Now, they can create virtual spectrometer in RADlab1.03 and use for training and practice. Adopting virtual spectrometer for training and practice will improve technologist's skill performing in vitro test. This will reduce the repeat rate of in vitro test and in this way it also has clinical significance.

CONCLUSION

The spectrum obtained with virtual spectrometer is in excellent agreement with theoretical and experimental results. The virtual spectrometer can be used for training purposes.