Evaluation of calibration factor of OSLD toward eye lens exposure dose measurement of medical staff during IVR

Abstract The eye lens is a sensitive organ of which an x‐ray exposure dose should be managed during interventional radiology (IVR). In the actual situations, the eye lens is exposed to scattered x‐rays; they have different from the standard x‐ray energies which are used for general dose calibration of the dosimeter. To perform precise dose measurement, the energy dependence of the dosimeter should be properly accounted for when calibrating the dosimeter. The vendor supplies a calibration factor using 80‐kV diagnostic x‐rays under a free‐air condition. However, whether it is possible to use this calibration factor to evaluate the air kerma during the evaluation of the eye lens dose is unclear. In this paper, we aim to precisely determine calibration factors, and also examine the possible application of using a vendor‐supplied calibration factor. First, the x‐ray spectrum at the eye lens position during fluoroscopy was measured with a CdTe x‐ray spectrometer. We mimicked transfemoral cardiac catheterization using a human‐type phantom. Second, we evaluated the doses and calibration factors at three dosimetric points: front and back of protective goggles, and the front of the head (eye lens position). We used the measured x‐ray spectrum to determine the incident photon distribution in the eye lens regions, and x‐ray spectra corresponding to the dosimetric points around the eye lens were estimated using Monte Carlo simulation. Although the calibration factors varied with dosimetric positions, we found that the factors obtained were similar to the vendor‐supplied calibration factor. Furthermore, based on the experiment, we propose a practical way to calibrate an OSL dosimeter in an actual clinical situation. A person evaluating doses can use a vendor‐supplied calibration factor without any corrections for energy dependences, only when they add a systematic uncertainty of 5%. This evidence will strongly support actual exposure dose measurement during a clinical study.


| INTRODUCTION
Currently, x-ray examinations are an essential technology for performing noninvasive medical diagnosis. During procedures, such as interventional radiology (IVR), medical staff are routinely exposed to scattered x-rays, 1-3 and many dose evaluators warn that medical doctors performing IVR procedures receive a large amount of x-ray exposure. For these doctors, reducing radiation-induced cataracts [4][5][6] should be a top priority, and thus in 2011, the International Commission on Radiation Protection (ICRP) recommended a new radiation dose limit of 20 mSv per year. 7,8 Based on this new recommendation, research concerning eye lens dosimetry has been very active. [9][10][11] In these studies, the eye lens doses were often evaluated based on indirect measurements using a personal dosimeter. Generally, a personal dosimeter can determine radiation type and energy by means of a special algorithm based on the differences of responses related to different radiation filters. Personal dosimeters focus on effective dose evaluation of the whole body, however, this is not suitable for directly evaluating the dose to a specific organ, such as the eye lens. In contrast, we focused our attention on the direct dose measurement of the eye lens precisely under actual conditions.
In order to achieve a direct dose measurement of the eye lens, a small-type OSL dosimeter (nanoDot TM , 10 mm × 10 mm × 2 mm t ) 12 is available. Although small-type OSL dosimeters do not have the ability to estimate both radiation type and energy, a person evaluating doses does not need information concerning radiation type because during IVR procedures medical staff are only exposed to xray photons. Thus, we need to only pay attention to the energy dependence 13,14 of the small-type OSL dosimeter. Basically speaking, the energy dependence of the dosimeter is the difference of mass energy-absorption coefficients between the OSL dosimeter and air 15 ; this means that energy dependence for polychromatic x-ray distribution should be analyzed with polychromatic x-rays and should not be treated as monochromatic x-rays. 16 The dose to the eye lens is mainly caused by scattered x-rays from the patient. 3,17 This means that a person evaluating doses should pay close attention to the verifications of the polychromatic x-ray spectrum related to the scattered and penetrating x-rays.
Moreover, for eye lens dose evaluation, the following special attention should be paid. During the current clinical diagnosis, medical doctors usually wear the protective goggles to reduce eye lens exposure. When eye lens dose is evaluated by a dosimeter, a person evaluating doses should consider the change in x-ray energy caused by the beam hardening effect 18 during the penetration of materials (goggles). Furthermore, effect of contamination from backscattering x-rays generated by the head of the operator has to be considered.
There are no previous studies in which these phenomena are taken into consideration for calibrating a dosimeter used to measure exposure dose to the eye lens.
Recently, we have proposed a precise calibration procedure in which we considered the polychromatic x-ray distribution and the energy dependence of an OSL dosimeter. 16 In this study, we applied this general method to a specific situation of eye lens dosimetry, and aimed to evaluate the precisely determined calibration factors. Because many clinical studies use a vendor-supplied calibration factor, we also investigated the applicability of this factor. The results derived from this study will play an important role in determining the usefulness of clinical data.
2 | MATERIALS AND METHODS 2.A | Calibration procedure taking into consideration energy dependence of the OSL dosimeter We will explain a procedure to determine dose calibration factors by taking into consideration the energy dependence of an OSL dosimeter and differences found in various x-ray spectra. Our previous study 16 aimed to derive a calibration factor used for general x-ray diagnosis, and the effects of beam hardening, scattered x-rays, and backscattering x-rays were evaluated. Namely, when an x-ray spectrum including the above effects is obtained, a proper calibration factor can be determined. In this study, we applied this procedure to eye lens dosimetry.
The following is a brief explanation of our calibration method.
The calibration factor of the OSL dosimeter is defined as the ratio of air kerma divided by the absorbed dose corresponding to the dosimeter. Then, the "calculated calibration factor (CCF)" can be determined using the following formula: Calculated calibration factor : CCF ¼ Air kerma Absorbed dose spectrum : (1) From a unique x-ray spectrum, both air kerma and absorbed dose can be calculated simultaneously. Then, air kerma and absorbed dose are calculated as follows: where C(E), E, and μ en (E)/ρ are the intensity of x-ray spectrum, energy, and mass energy-absorption coefficient of air, respectively.
Eff.(E) is the energy dependence of the OSL dosimeter. 13 It is important that we determine a calibration factor when deriving the x-ray spectrum. The calibration factor described in Eq. (1) can provide an absolute value, but in a practical case, only a measurable value (the response of the OSL dosimeter: counts), which is proportional to the absorbed dose, can be obtained. In actual analysis when using an OSL dosimeter reading device, relative responses can be read out. In order to analyze the relationship between an absolute value and a relative value, we need to perform additional experiments to derive an actual calibration factor. Here, the vendor-supplied calibration factor used for a commercially available OSL dosimeter (nanoDot TM , Landauer, Inc., Illinois, USA) was determined using 80-kV diagnostic x-rays with a half-value layer (HVL) of 3.01 mm aluminum; this quality of x-ray is known as an RQR6 beam. 19 Using this radiation qual- (4) where "Counts" and "ϵ" are the response of the OSL dosimeter and intrinsic detection efficiency for each dosimeter, respectively. An experiment for the determination of the conversion coefficient f will be described later.
2.B | Estimation of x-ray spectra at dosimetric points We will describe an experiment and a simulation used to obtain the x-ray spectra around the eye lens position. First, we performed a phantom experiment as shown in Fig. 1(a); clinical application of an IVR procedure was mimicked using a human body equivalent phantom (PBU-60, Kyoto Kagaku Co., Ltd., Kyoto, Japan) as a patient. to obtain x-ray spectrum were appropriately applied.
In order to estimate the effects of x-ray attenuation in protective goggles and the effect of contamination of backscattering x-rays from the head of the operator, a Monte Carlo simulation was performed using EGS5 (electron-gamma-shower version 5) code. 21  2.C | Experiment to determine the actual calibration factor In this section, we will describe the experimental procedure to determine the conversion coefficient f in Eq. (4). Because the conversion coefficient f is only effected by the counting efficiency of the reading device, it can be determined from the experiment, in which x-ray exposure to the OSL dosimeter was performed using diagnostic xray equipment.
The experimental value of Counts/ϵ and air kerma was obtained using an OSL dosimeter and an ionization chamber, respectively. Diagnostic x-ray equipment (MRAD-A 50S/70, Canon Medical Systems Corp., Tochigi, Japan) was used. In order to reduce contamination from scattered x-rays generated from the movable diaphragm of the x-ray equipment, the OSL dosimeter and ionization chamber were individually placed in a lead shielding box having a window opening of 100mm ϕ and an additional lead collimator (25mm ϕ ) was set in front of the x-ray equipment. 23 The distance between the x-ray focal point and the dosimeter was set at 200 cm. Irradiation conditions were 80 kV (tungsten target with total filtration of 2.5 mm aluminum), 100 mA, and 2 s. The response of the OSL dosimeter was defined as Counts/ϵ; where "Counts" and "ϵ" were measured response and intrinsic detection sensitivity, respectively. 13,16 To reduce statistical uncertainty, three OSL dosimeters were individually irradiated at the above conditions, and each dosimeter was read five times using a commercially available reading device (microStar, Landauer, Inc., Illinois, USA). We adopted the mean value of Counts/ϵ for 15 readings. The signal loss caused by multiple readings was corrected. 12,24 Here, we used an "unscreened"-type nanoDot OSL dosimeter, and we determined the accuracy related to the sensitivity of each nanoDot OSL dosimeter.
According to previous research, 25  pressure of 1013 hPa. 27 The air kerma was measured five times, and the mean value of air kerma was adopted. The mean value of Counts/ϵ was approximately 14,500 counts, and the averaged value of air kerma was determined to be approximately 3 mGy. Using the response of the OSL dosimeter and air kerma, the conversion coefficient f in Eq. (4) was calculated. Then, the correlation between the calibration factor and effective energy of x-ray spectra was examined. When the x-ray spectrum is obtained, the corresponding air kerma can be calculated.
Additionally, the calculation procedure for attenuation related to aluminum is well known. Combining this knowledge, the air kerma corresponding to the x-ray spectrum after penetrating the aluminum having thickness t can be described as follows: where C(E), μ en E ð Þ=ρ ð Þ air and μ(E) Al are the intensity of measured xray spectrum, the mass energy-absorption coefficient of air and the linear attenuation coefficient of aluminum, respectively. 28 In order to obtain the effective energy, a virtual experiment was performed and the half-value layer (HVL) was obtained from the attenuation curve using air kerma. Here, since the HVL is the amount of attenuation corresponding to the effective energy of the continuous x-ray spectrum, the effective linear attenuation coefficient μ eff can be derived from the following relationship: Then, because μ has a unique relationship with energy E for monoenergetic x-ray, 28 μ eff obtained from the above equation can be converted to effective energy using this relationship.

2.D | Demonstration of eye lens dosimetry during fluoroscopic examination
In order to demonstrate the applicability of our calibration procedure, we performed a phantom study using an OSL dosimeter. Figure 2 shows photographs of the experiment.  Here, we will demonstrate the eye lens dosimetry based on precisely determined calibration factors. Figure 4 shows the results of the experiment for eye lens dosimetry during fluoroscopic examination. We measured the eye lens dose using small-type OSL dosimeters. The experimental arrangement is shown in Fig. 2, and the radiation dose was analyzed using our method. In our experiment, a human dummy consisting of expanded polystyrene was used instead of an actual operator, therefore the effect of backscattering x-rays from the human head on the measured dose should be additionally estimated. In order to estimate the effect of the backscattering x-rays, we analyzed the contribution of the effect using Monte Carlo simulation. Table 1

| DISCUSSION
In this paper, we determined a calibration factor for a small-type OSL dosimeter in order to achieve a precise evaluation of the eye lens dose during fluoroscopic examination. By considering the variation of x-ray spectra under actual conditions and the energy dependence of the OSL dosimeter, the calibration factors of outside and inside protective goggles, and the eye lens position were determined. We found it is very meaningful to determine the calibration factors precisely for each place. We also found the usefulness of the calibration factor provided by the vendor. We will discuss the details below.

4.A | Usefulness of vendor-supplied calibration factor for eye lens dosimetry
We will discuss the calibration factors of the OSL dosimeter determined in this study. We found it interesting that the calibration factors concerning eye lens dosimetry were similar to the vendorsupplied calibration factor as shown in Fig. 3(c).  Recently Tanaka et al. reported the measured dose of scattered xrays during IVR procedure, and in their study dose calibration was performed using 80 kV x-rays. 30 In our paper, we verified that the calibration factor determined using standard x-rays (80-kV direct xrays) can be applied to measure the dose during IVR, therefore our result justifies the previously measured data. Here, the standard calibration factor in this study was derived using 80 kV x-rays (effective It is known that OSL dosimeters have relatively large energy dependence in the lower energy region, [13][14][15] and it is considered that calibration factors used for eye lens dosimetry during IVR procedures should be determined as described above. However, not all dose evaluators can use the above-mentioned procedure in which the energy dependence of the OSL dosimeter is carefully considered.
Here, we propose a practical procedure based on our results as follows. The results in Fig. 3 indicate that the difference between a vendor-supplied calibration factor and the factors determined in this study are within 5%. This means that the vendor-supplied calibration factor can be applied to the dose evaluation of eye lens during an IVR procedure under the following limitation; a person evaluating doses should add a systematic uncertainty of 5%. This is a valuable conclusion that was derived based on the research explained in this paper. In our experiment, the tube voltage of 74 kV determined by the automatic exposure control system was used. When different xray qualities are used, the energy dependence of the OSL dosimeter needs to be considered. As mentioned above, because the energy dependence of the OSL dosimeter is large, [13][14][15] we need to pay special attention to this phenomenon.
In this study, we focused our attention on energy dependence as it affects the calibration factor, and we did not consider the effect of angular dependence. Even if x-rays are incident on the dosimeter obliquely, there is almost no change in the sensitivity, but when x-rays enter the dosimeter from the lateral direction, a very large decrease in sensitivity is observed. This effect was known as "angular dependence". 25,31,32 In clinical cases, it is diffi- We will discuss the eye lens doses measured using an OSL dosimeter in the phantom experiment. Using precise calibration factors determined in Fig. 3(c), we evaluated the effect of protective goggles on dose reduction. As shown in Fig. 4, the presence of protective goggles was clearly observed. In our experiment, the reduction rate of exposure dose using protective goggles was analyzed to be 70%. However, basically speaking, the reduction rate using the protective goggles depends on the design of the goggles and analytical procedure. 33,34 Even though a 70% reduction was one of the experimental results, we recommend wearing goggles when performing IVR procedure. In actual situations, the radiation dosimeter may not be attached directly to an actual eye lens. In this case, the eye lens dose may be inferred from the data at other dosimetric positions; for example, the outside and inside positions of the protective goggles are candidates. This is a limitation of this evaluation. In our experiment, we used a human dummy in order to imitate actual body positions during an IVR procedure. Contribution of backscattering x-rays was not included because the human dummy does not consist of human equivalent material.
Based on our simulation result, the contribution of backscattering x-rays is estimated to be 35% at eye lens position and that corrected data are presented in Fig. 4. We expect that accurate clinical data will be obtained with precise calibration using our data.
The ICRP recommended to use operational quantity for monitoring eye lens dose using a dose equivalent at 3 mm depth: H p (3) 35,36 instead of air kerma. We will describe the difference between air kerma and H p (3). The H p (3) is calculated using the following formula: where h pK (3) is the conversion coefficient from air kerma to H p (3).
The h pK (3) depends on the energy and the angle of incident x-rays.
In addition, the size and shape of the calibration phantoms are important factors in the determination of h pK (3). 37 There have been many studies on these investigations. International standards of IEC 62387-2012, ISO 4037-1, and 12789-2 referred the conversion coefficient at 3-mm depth using a slab phantom made of acrylic or soft-tissue equivalent material for calibration. [38][39][40] Moreover, the shape of the phantom has been reported elsewhere. 29,37,41 In this paper, we reported a precise calibration procedure of measuring "air kerma." In order to evaluate eye lens dose, air kerma should be con- We expected that the evidence and procedure play an important role for obtaining accurate value using OSL dosimeter.

| CONCLUSION S
In order to achieve precise dose evaluation of the eye lens of an operator when using an OSL dosimeter for the actual clinical measurement, we determined calibration factors. Our method took into consideration the difference in x-ray spectra and the energy dependence of the dosimeter. Calibration factors at three dosimetric positions were evaluated: front and inside of the protective goggles, and eye lens. We found that the values of calibration factors varied for vendor-supplied factor which was determined using 80 kV x-rays. In the case that a person evaluating doses can apply our results, the exposure doses can be determined precisely. On the other hand, when they cannot apply our results because of restrictions related to actual clinical situations, we proposed a practical way. A person evaluating doses can use the vendor-supplied calibration factor with a systematic uncertainty of 5%. Note that this conclusion can be applied to a fluoroscopic system generating 74 kV x-rays (determined by an automatic exposure control system). Strictly speaking, when we want to evaluate doses using other x-ray equipment in which different x-ray qualities are applied, energy dependence should be accounted for.

ACKNOWLEDG MENTS
The authors thank Mr. Kenji Yamada of Tokushima University Hospital, Japan for technical assistance with the experiments.

CONFLI CT OF INTEREST
The authors of T. Okazaki