Characterization and implementation of OSL dosimeters for use in evaluating the efficacy of organ-based tube current modulation for CT scans of the face and orbits R. M. Marsh and M. Siloskya) Department of Radiology, University of Colorado School of Medicine, Aurora, Colorado 80045

(Received 16 September 2014; revised 13 January 2015; accepted for publication 18 February 2015; published 18 March 2015) Purpose: The purpose of this work was to characterize commercially available optically stimulated luminescent (OSL) dosimeters for general clinical applications and apply the results to the development of a method to evaluate the efficacy of a vendor-specific organ-based tube current modulation application for both phantom and clinical computed tomography (CT) scans of the face and orbits. Methods: This study consisted of three components: (1) thorough characterization of the dosimeters for CT scans in phantom, including evaluations of depletion, fading, angular dependence, and conversion from counts to absorbed dose; (2) evaluation of the efficacy of using plastic glasses to position the dosimeters over the eyes in both phantom and clinical studies; and (3) preliminary dosimetry measurements made using organ-based tube current modulation in computed tomography dose index (CTDI) and anthropomorphic phantom studies. Results: (1) Depletion effects were found to have a linear relationship with the output of the OSL dosimeters (R2 = 0.96). Fading was found to affect dosimeter readings during the first two hours following exposure but had no effect during the remaining 60-h period observed. No significant angular dependence was observed for the exposure conditions used in this study (with p-values ranging from 0.9 to 0.26 for all t-tests). Dosimeter counts varied linearly with absorbed dose when measured in the center and 12 o’clock positions of the CTDI phantoms. These linear models of counts versus absorbed dose had overlapping 95% confidence intervals for the intercepts but not for the slopes. (2) When dosimeters were positioned using safety glasses, there was no adverse effect on image quality, and there was no statistically significant difference between this placement and placement of the dosimeters directly on the eyes of the phantom (p = 0.24). (3) When using organ-based tube current modulation, the dose to the lens of the eye was reduced between 19% and 43%, depending on the scan protocol used and the positioning of the phantom. Furthermore, the amount of dose reduction was significantly affected by the vertical position of the phantom, with the largest reduction in dose seen when the phantom was centered in the gantry. Conclusions: (1) An appropriate correction factor, specific to CT scanning, was developed to account for depletion and fading characteristics of the dosimeters. Additionally, an equation to convert dosimeter counts to absorbed dose was established. (2) The use of plastic safety glasses was validated as an appropriate positioning device when measuring dose to the lens of the eye. (3) The use of organ-based tube current modulation can reduce dose to the lens of the eye during CT scanning. The amount of dose reduction, however, is largely influenced by the positioning of the anatomy in the gantry. C 2015 American Association of Physicists in Medicine. [http://dx.doi.org/10.1118/1.4915076] Key words: CT dose, OSL dosimetry, X-CARE, face and orbits, dose optimization 1. INTRODUCTION 1.A. Background

In recent years, manufacturers of computed tomography (CT) scanners have developed a number of technologies designed to minimize dose to radiosensitive tissues, quoting varying degrees of dose reduction. One such technology is organbased tube current modulation where the tube current is reduced over the anterior portion of the patient with the intent of reducing dose to radiosensitive structures such as the lens of the eyes, thyroid, and breasts. The tube current is then increased on the posterior side of the patient in order to maintain image quality. Siemens Medical Systems (Malvern, PA) claims dose reduction of up to 40% for 1730

Med. Phys. 42 (4), April 2015

their organ-based tube current modulation (X-CARE).1 To validate these claims, a number of phantom studies have been performed reporting varying levels of dose reduction.2,3 However, clinical studies to evaluate the efficacy of this product in reducing dose to the lens of the eye during patient exams remain difficult to perform. The purpose of this study was to develop and evaluate a method for performing measurements of dose to the lens of the eye using nanoDot optically stimulated luminescent (OSL) dosimeters (Landauer, Glenwood, IL). Although others have characterized nanoDot performance for CT x-ray beams, these studies have been performed in air4 or with the x-ray tube positioned at a fixed location rather than exposing the nanoDots to a full 360◦ tube rotation.5

0094-2405/2015/42(4)/1730/9/$30.00

© 2015 Am. Assoc. Phys. Med.

1730

1731

R. M. Marsh and M. Silosky: Organ-based mA modulation for face and orbits scans

The first goal of this study was to thoroughly characterize the nanoDot dosimeters under realistic scan conditions. This set the basis for establishing a conversion from counts to absorbed dose and an estimate of error associated with subsequent measurements by evaluating the effects of depletion (loss of signal due to multiple readings), fading (loss of signal due to differences in time between exposure and reading), angular dependence of the nanoDot dosimeters in CT dose measurements, and the determination of an equation for the conversion of nanoDot counts to absorbed dose and the error associated with these values. Second, a method for implementing the use of nanoDots in clinical studies where they will be placed near a patient’s eyes was developed. Finally, a phantom study was performed to simulate clinical conditions and establish the basis for a clinical study evaluating the efficacy of X-CARE in reducing dose to the lens of the eye for exams of the face and orbits, a common clinical exam that routinely includes the entire lens in the scanned volume.

2. METHODS All experiments were performed using a Siemens Definition Flash CT scanner. Two types of dosimeters were used— nanoDot OSL dosimeters and a 0.6 cm3 ion chamber (Radcal, Monrovia, CA). The nanoDots consist of a 4 mm-diameter Al2O3:C disk enclosed in a 1 cm × 1 cm × 2 mm plastic case.6 The nanoDots used in this study have a reported precision of 5%. To account for small variations in manufacturing and material characteristics, the manufacturer specifies a sensitivity factor for each nanoDot. This sensitivity factor is a constant, applied to correct the number of measured counts, as shown in (counts) , (1) CS = S where CS is the sensitivity-corrected counts and S is the manufacturer-defined sensitivity factor (S ∈ [0.85,0.97]). This sensitivity factor was applied to all nanoDot measurements. All ion chamber measurements were reported as air kerma, displayed in mGy. Two phantoms were used throughout this study: a 16 cmdiameter computed tomography dose index (CTDI) head phantom (Fluke Biomedical, Everett, WA) and an anthropomorphic head phantom (Radiology Support Devices, Inc., Long Beach, CA). The location of the dosimeter within the CTDI phantom is designated as follows: CTDIcenter indicates that the dosimeter was placed in the center hole of the phantom and CTDI12 indicates placement in the hole at the 12 o’clock position. The scans described in Secs. 2.A–2.D were performed in axial mode with the following acquisition parameters: 120 kVp, 64 × 0.6 mm detector configuration, and 1 s rotation time. The mA s was varied and is indicated, as appropriate, in each section. The scans described in Secs. 2.E–2.G were based on two clinical protocols used at our institution: a routine face and orbits protocol and a routine head protocol. The routine face and orbits protocol had the following scan parameters: Medical Physics, Vol. 42, No. 4, April 2015

1731

120 kVp, 1 s rotation time, pitch = 0.8, reference mA s = 55. The routine head protocol had parameters of 120 kVp, 1 s rotation time, axial acquisition, reference mA s = 340. The beam width for both protocols was 38.4 mm. To simulate clinical practice of scans of the face and orbits, the scan length for the anthropomorphic head phantom went from the tip of the mandible through the frontal sinuses. 2.A. Evaluation of nanoDot depletion

While OSL dosimeters can be read multiple times following a single irradiation, some signal loss occurs with each read. This signal loss is known as depletion. While others have investigated the effects of depletion,4,7 some of these studies were performed using modified (i.e., noncommercially available) OSL holders and/or readers.8–10 This may account for the variation seen in depletion and fading effects. Others characterized OSL dosimeters at much higher photon energies (6–15 MV) or electron energies.7,11 While some others have characterized commercially available nanoDots at diagnostic energies,4,12 we sought to independently evaluate these effects to determine if these results were generally applicable or relevant only for a specific batch of nanoDots. It has been previously demonstrated that at diagnostic energies, depletion kinetics of OSL dosimeters are not energydependent.13 Consequently, a single experimental setup was used. To evaluate the effects of depletion on multiple nanoDot readings, a custom positioning device (12 cm polystyrene cylinder with slots cut through the center) was designed such that two nanoDots could be reproducibly positioned in the CTDI phantom. The positioning device was placed at CTDIcenter such that the dots were centered in the phantom as illustrated in Figs. 1(A) and 1(B). The phantom was then centered in the CT scanner and an axial scan was performed with a detector configuration (64 × 0.6 mm) such that both nanoDots were well within the edges of the primary beam. The resulting air kerma at the center of the phantom was approximately 25 mGy. This procedure was repeated three times, resulting in a total of six nanoDots scanned. After approximately 24 h, each nanoDot was read out 20 consecutive times using a Landauer Microstar reader. This

F. 1. A 16 cm CTDI phantom (A) with the custom made nanoDot positioning device inserted in the center hole. Placement of the nanoDots within the positioning device (B) clearly falls within the axial beam extent, represented by the dashed lines.

1732

R. M. Marsh and M. Silosky: Organ-based mA modulation for face and orbits scans

measurement time point was chosen to ensure the depletion measurements were not affected by fading (described below in Sec. 2.B). The timing of the reading was also conducive for clinical implementation of nanoDot measurements. The mean sensitivity-corrected counts (CS ), standard deviation, and coefficient of variation (COV) were calculated for each reading. Cs was plotted as a function of the number of reads, normalized to the first reading, and fit via linear regression to yield an equation describing the depletion effect. This relationship and its associated error were used to correct for multiple readings taken for all additional experiments. 2.B. Evaluation of nanoDot fading

The nanoDots signal decreases, or fades, over time following irradiation. While fading has been evaluated by other investigators, the effects were studied here in diagnostic energy ranges and with commercially available equipment, for many of the reasons mentioned in Sec. 2.A. To evaluate the effects of fading, five pairs of nanoDots were scanned using the positioning device and CTDI phantom as described in Sec. 2.A. The nanoDots were positioned at CTDIcenter and exposed to an air kerma of approximately 40 mGy. The dosimeters were read 20 min after scanning and reread approximately every two hours for a period of 36 h. Less frequent readings were then performed until 60 h postexposure. The mean sensitivity- and depletion-corrected counts (CSD) was calculated for each time point and plotted as a function of the time between exposure and reading. An estimate of the error in these values was performed based on the error in the depletion correction (Sec. 3.A) and the standard deviation of the sensitivity-corrected counts. 2.C. Evaluation of angular dependence

Although nanoDots are treated as point dosimeters, they have a finite size and nonspherical shape. Consequently, it was important to characterize the effect of angular positioning on dose response. Others have found no angular dependence for axial CT scans for measurements made in air.4,12 Lavoie et al.5 previously looked at the angular dependence by irradiating nanoDots with the CT x-ray tube fixed at set positions between 0◦ and 360◦ in increments of 10◦. They found no dependence on x-ray tube angle for measurements made in phantom, but their evaluations were limited to dosimeters placed at scanner isocenter. In the case where the dosimeters are exposed under radially symmetric conditions (i.e., with surrounding materials and placement such that the radiation reaching the dosimeter is the same from all angles), no angular dependence is expected. However, previous studies have not evaluated the situation where, due to varying material thicknesses or off-isocenter positioning, the dosimeters are exposed to a nonradially symmetric radiation distribution. To address this concern, angular dependence was evaluated at both CTDIcenter and CTDI12. Fifteen nanoDots were scanned (one at a time) using the setup described in Sec. 2.A, with the nanoDots placed at CTDIcenter. Five nanoDots were scanned in each of the Medical Physics, Vol. 42, No. 4, April 2015

1732

following positions: with the labels facing upward (0◦), with the labels facing right (+90◦), and with the labels facing left (−90◦). These measurements were then repeated with the nanoDots placed at CTDI12. Use of the CTDI phantom allowed for accurate, reproducible positioning. The nanoDots were read 24 h after scanning and the mean value, standard deviation, and coefficient of variation were determined for each group of dosimeters. t-tests were performed to determine if angular positioning had a statistically significant effect on nanoDot response. 2.D. Conversion from counts to absorbed dose

An individual nanoDot read is displayed as the number of counts registered by the Microstar reader. In order to convert from counts to absorbed dose, the relationship between these quantities was characterized for the scan conditions used in this study. A previous study demonstrated that the conversion factor from counts to absorbed dose is energydependent, but this dependence was only characterized for a single dose and did not address whether a single linear conversion factor was appropriate used across a range of doses.5 Furthermore, previous studies of relationship between counts and absorbed dose have been performed relative to the manufacturer’s calibrations performed at 80 kVp.16 This study sought to establish a conversion factor that is independent of reader calibrations performed by the manufacturer. Knowing that there are differences in the spectrum of the xray beam at different locations in the phantom, two trials were performed to determine if a single conversion from counts to absorbed dose could be utilized for future studies. The first trial used the experimental setup illustrated in Figs. 1(A) and 1(B), where the dosimeters were placed at CTDIcenter. Two adjacent nanoDots were placed at CTDIcenter and an axial scan was performed as described in Sec. 2.A. This procedure was repeated twice resulting in six exposed nanoDot dosimeters. The nanoDots were then removed and a 0.6 cm3 ion chamber was placed at CTDIcenter. The scan was repeated three times, yielding three ion chamber measurements. The tube current was then increased and these steps were repeated. This entire procedure was repeated, incrementally increasing mA s from 40 to 550, corresponding ion chamber dose measurements between 3.48 and 43.39 mGy. It should be noted that the CTDI phantom was chosen for this experiment as opposed to an anthropomorphic phantom due to the need to accurately and reproducibly position the 0.6 cm3 ion chamber. The second trial was performed in a similar manner but with the dosimeters placed at CTDI12, with a corresponding ion chamber measurement between 4.77 and 62.58 mGy. The purpose was to determine if there was a statistically significant difference in the conversion from counts-to-dose for different locations in the phantom and to identify an appropriate conversion for nanoDots affixed near a patient’s eyes. After 24 h, the nanoDots were each read three times. The raw counts from the nanoDot readings were corrected for sensitivity and depletion and the mean nanoDot readings, mean ion chamber readings, and their respective standard deviations were calculated for each tube current selection. The

1733

R. M. Marsh and M. Silosky: Organ-based mA modulation for face and orbits scans

mean ion chamber reading was converted to dose to tissue by multiplying by 1.06 as recommended in AAPM Report 96.14 CSD was plotted as a function of the mean dose to tissue and fit via linear regression with the 95% confidence intervals calculated for the slope and intercept for each position in the phantom. A conversion from nanoDot counts to absorbed dose to tissue was identified based on this regression. 2.E. Evaluation of nanoDot positioning method for phantom studies

Because the ultimate goal of this work was to measure dose to the lens of the eye in clinical studies, it was necessary to identify a reproducible method for positioning the nanoDots that would not interfere with the measurements or adversely affect image quality. It was also necessary that the method met our institution’s infection control standards and did not place an undue burden on clinical staff. To work within these constraints, thin plastic safety glasses were purchased, and all metal components were removed. The glasses could be worn by a patient or placed on an anthropomorphic head phantom. Furthermore, dosimeters could be positioned on the outside of these glasses in a reproducible location, allowing for both convenient measurements and the ability to sterilize the positioning device (i.e., glasses) between patients as shown in Fig. 2. A qualitative image evaluation was performed by a neuroradiologist and determined that the modified safety glasses, though visible in the images, did not adversely affect image quality. To determine if the use of safety glasses would substantially affect dose measurements, the anthropomorphic head phantom was scanned using our institution’s standard face and orbits protocol with nanoDots placed directly on the left and right eyes. Three repeat measurements were preformed, resulting in a total of eight exposed nanoDots. The procedure was then repeated with the nanoDots taped to the outer surface of the safety glasses, centered over the left and right eyes. After

1733

24 h, each nanoDot was read three times. The readings were corrected for sensitivity and depletion and the mean value and its associated error were calculated for each setup. A t-test was performed to determine if there was a statistically significant difference in the mean values and to determine the effect the presence of the glasses had on dose measurements. 2.F. Effect of vertical position on dose reduction using X-CARE

While a patient’s head is typically centered in the gantry for many clinical exams, it is common practice at our facility to lower the table height during exams of the face and orbits so that the anatomy of interest is at isocenter. However, XCARE is designed assuming that the patient is centered in the gantry. To evaluate X-CARE as a dose reduction technique for our clinical practice, several experiments were performed to investigate the effect of vertical position and to provide an estimate of the dose reduction to be expected during patient scans. For the first experiment, three nanoDots were placed on the anterior side of the CTDI phantom. This established a generalized assessment of the effect of vertical positioning on the efficacy of X-CARE, without the added complexities of phantom structure or radial asymmetry. The phantom was positioned in the center of the CT gantry and scanned using the standard face and orbits protocol, without X-CARE. This process was repeated four times, lowering the table (and the phantom) 2 cm each time, resulting in a maximum difference in vertical position of 8 cm. The entire procedure was performed three additional times, once using the equivalent face and orbits protocol with X-CARE, once using the routine head protocol without X-CARE, and once using the routine head protocol with X-CARE. After 24 h, each nanoDot was read three times. The readings were corrected for sensitivity and depletion. The mean value and associated error were calculated for each vertical position and the percent difference between X-CARE and routine (non-X-CARE) protocols was plotted as a function of vertical position, relative to scanner isocenter. 2.G. Anthropomorphic phantom study

F. 2. An anthropomorphic head phantom is shown with nanoDot OSL dosimeters affixed to plastic safety goggles and placed over the eyes of the phantom. The intent was to employ the same positioning method in a clinical study. Medical Physics, Vol. 42, No. 4, April 2015

The second experiment characterized the effect of XCARE using a phantom that was more representative of human anatomy. Because different behaviors were observed for routine head and face and orbits protocols and at different vertical positions (Sec. 3.F), a number of different scans were performed. An anthropomorphic head phantom was scanned with nanoDots affixed over the left and right eyes using the safety glasses described in Sec. 2.E. Eight different scan conditions were evaluated including protocols for both face and orbits and routine head, with and without X-CARE, and at two vertical positions: with the phantom centered in the gantry and with the orbits centered in the gantry. For each condition, five scans were performed for a total of ten exposed dosimeters. After 24 h, each nanoDot was read three times. The readings were corrected for sensitivity and depletion and

1734

R. M. Marsh and M. Silosky: Organ-based mA modulation for face and orbits scans

1734

the mean value and associated error were calculated. The relative dose (normalized to the highest dose scan condition) was calculated for each scan condition.

3. RESULTS 3.A. Evaluation of nanoDot depletion

The relationship between the number of times the nanoDots were read and the relative mean counts (normalized to the first read) is shown in Fig. 3. CS decreased linearly with the number of reads with a slope of −0.0026 ± 0.0001, an intercept of 0.9944 ± 0.0016, and an R2 value of 0.96. Over twenty readings, CS was reduced to 94.5% of its initial value. The COV varied between 2.2% and 3.0% with no trend observed. Based on these results, the following equation was used to correct consecutive nanoDot readings for the effects of depletion: CSD = CS ×

1 , (−0.0026 × n + 0.9944)

(2)

where CSD is the sensitivity- and depletion-corrected counts, CS is the sensitivity-corrected counts, and n is the number of times the nanoDot has been read following irradiation.

F. 4. Normalized nanoDot readings are plotted as a function of time after exposure. A substantial drop in counts was observed during the first two hours after exposure. Following this, readings were relatively stable with variations in mean counts well within the statistical error in the measurements. For consistency, all readings in subsequent experiments were performed approximately 24 h following exposure.

p = 0.57). Additionally, no statistically significant difference was observed when comparing the readings between angles for the 12 o’clock position (−90◦ to 0◦, p = 0.26, +90◦ to 0◦, p = 0.78, −90◦ to +90◦, p = 0.56). 3.D. Conversion from counts to absorbed dose

3.B. Evaluation of nanoDot fading

Figure 4 shows the effect of fading on nanoDot readings, with the relative mean counts (normalized to the first read) plotted as a function of the time elapsed between exposure and reading. CSD decreased within the first two hours following exposure. However, beyond 2 h after exposure, CSD varied by less than 2.3%, which was within the error for these measurements, as shown by the error bars in Fig. 4. 3.C. Evaluation of angular dependence

Table I lists CSD, standard deviation, and COV for nanoDots scanned at each combination of phantom position and angle. No statistically significant difference was observed when comparing the readings between angles for the center position (−90◦ to 0◦, p = 0.90, +90◦ to 0◦, p = 0.48, −90◦ to +90◦,

Figure 5 shows a plot of CSD as a function of the absorbed dose to tissue for both scanners. The data for both the center and 12 o’clock positions were well fit using linear regression with values of R2 > 0.99 for both datasets. Table II lists the slopes, intercepts, and respective 95% confidence intervals for both positions. There was overlap in the 95% confidence intervals of the intercepts while there was no overlap in the confidence intervals of the slopes. As a result, the difference in the regression lines becomes more substantial as the measured counts increase. Consequently, we chose to use the data obtained from the 12 o’clock position to establish a counts-to-dose conversion factor, since that position more accurately represents the position of the eyes and hence the conditions of the upcoming clinical study. However, at the low doses seen in this study, there was no statistical difference in the conversion factors found at the center and the 12 o’clock positions, demonstrating that at low doses, differences in scatter play an insignificant

T I. The mean sensitivity- and depletion-corrected counts (C SD) standard deviation, and COV for each angle and position are tabulated. No statistically significant difference was observed when comparing mean values between angles. Position

Center F. 3. Data illustrating the effect of depletion are shown with the normalized mean counts plotted as a function of the number of reads. The data are well-modeled using linear regression, with an R 2 = 0.96. An equation to correct consecutive reads for depletion was determined based on this analysis. Medical Physics, Vol. 42, No. 4, April 2015

12 o’clock

Angle

(C SD)

Standard deviation

COV

−90 0 90 −90 0 90

48 868.3 48 224.3 48 418.8 65 377.3 65 527.1 64 664.5

961.7 687.3 1 364.8 1 885.1 1 779.7 1 876.0

1.97 1.43 2.82 2.88 2.72 2.90

1735

R. M. Marsh and M. Silosky: Organ-based mA modulation for face and orbits scans

F. 5. The absorbed dose to tissue was plotted as a function of sensitivityand depletion-corrected mean counts for both the center and 12 o’clock positions. Both datasets were well fit using linear regression with the dashes representing the regression lines and the solid gray and black lines representing the 95% confidence intervals for the center and 12 o’clock position, respectively. It should be noted that at low doses, there is substantial overlap between confidence intervals.

role in the count-to-dose conversion factor. Equation (3) is the conversion equation, where D is the absorbed dose to tissue and CSD is the mean sensitivity- and depletion-corrected counts, D(mGy) = CSD × 5.972 (±0.024) × 10−4 − 0.1456 (±0.1383).

(3)

3.E. Evaluation of nanoDot positioning method for phantom studies

When the nanoDots were positioned directly on the phantom, CSD was 11 608 ± 740 counts (COV = 6.4%). With the nanoDots positioned on the outside of the safety glasses, CSD was 10 991±646 counts (COV = 5.9%). The difference in the means was not statistically significant (p = 0.24). 3.F. Effect of vertical position on dose reduction using X-CARE

Figure 6 shows the difference in CSD between scans performed with and without X-CARE, expressed as the percentage of CSD for the non-X-CARE scans. Data are shown as a function of vertical distance from isocenter for both the head and face and orbits protocols. With the center of the phantom positioned at isocenter, X-CARE provided a dose

1735

F. 6. The percent dose reduction resulting from the use of X-CARE, compared to the routine, non-X-CARE protocols, is plotted as a function of the distance of the center of the phantom from isocenter. While lowering the phantom led to a reduction in the efficacy of X-CARE, trends in the data are difficult to characterize. Using X-CARE in conjunction with lowering the phantom 8 cm resulted in a dose reduction of 21.1% and 19.7% for the head and face and orbits protocols, respectively. This is in comparison to more substantial dose reductions of 28.8% and 33.9% when using X-CARE and positioning the center of the phantom at isocenter.

reduction of 28.8% for the face and orbits protocol and a 33.9% dose reduction for the head protocol. At positions below isocenter, the dose savings were reduced to between 16.5% and 19.6% for the face and orbits protocol and to between 21.1% and 29.3% for head protocol. While lowering the phantom clearly led to reduced dose savings, it is difficult to characterize the trend based on the measurements available. 3.G. Anthropomorphic phantom study

Figure 7 shows the dose measured for each scan condition described in Sec. 2.G, normalized to the maximum dose measured for each protocol. For both protocols, the largest dose to the lens of the eye was seen when the orbits were positioned at isocenter and X-CARE was not used. The lowest dose was achieved when the center of the phantom was positioned at isocenter and X-CARE was used. Under these circumstances, the relative dose was 61% and 57%, respectively, for the face and orbits and head protocols. Using X-CARE with the orbits positioned at isocenter led to similar reduction for both scan protocols with relative doses of 82% and 81% for the face and orbits protocol and the head protocol, respectively. Positioning the center of the phantom at isocenter, with X-CARE turned off, was the only condition tested in which there was a substantial difference in relative dose between the two protocols.

T II. The slopes, intercepts, and 95% confidence intervals for the linear fits (shown in Fig. 6) are shown. The confidence intervals of the intercepts overlap but the confidence intervals of the slopes do not. Position Center 12 o’clock

Slope

95% confidence interval

Intercept

95% confidence interval

5.667 × 10−4 5.972 × 10−4

[5.617 × 10−4, 5.716 × 10−4] [5.911 × 10−4, 6.032 × 10−4]

0.1676 −0.1456

[−0.0431, 0.3782] [−0.5020, 0.2091]

Medical Physics, Vol. 42, No. 4, April 2015

1736

R. M. Marsh and M. Silosky: Organ-based mA modulation for face and orbits scans

1736

measurements over the next 58 h. Other studies have shown varying degrees of fading effects. One study, performed at therapeutic photon energies, showed as much as 38% signal loss due to fading,7 while another saw fading effects similar to those reported here,12 although fading was only measured over a 60-min time period. As a matter of logistical simplicity, all subsequent measurements occurred approximately 24 h following initial exposure. However, no substantial fading effect was observed after the first two hours. For the upcoming clinical trial, it is recommended that a minimum of two hours elapse between nanoDot exposure and reading. F. 7. The measured dose, normalized to the highest dose scan condition (orbits centered and X-CARE turned off), is shown for the four conditions tested for both face and orbits and head protocols. Using X-CARE, while keeping the phantom in the orbits centered position, reduced dose by approximately 20% for both protocols. Keeping X-CARE off, raising the phantom from the orbits centered position to the head centered position reduced dose by 36% for the face and orbits protocol and 20% for the head protocol. Positioning the center of the phantom at isocenter and using X-CARE led to an additional reduction for the head protocol but only marginal improvement for the face and orbits protocol.

4. DISCUSSION 4.A. Correction of nanoDot readings for depletion and fading

As shown in Fig. 3, mean nanoDot counts decreased linearly with the number of readings. The observed effect was small with only a 5.5% decrease over twenty readings. However, given the strength of the trend, a correction factor was applied to all nanoDot readings for the remainder of the experiments (evaluation of fading, angular dependence, conversion to from counts-to-dose, the effect of positioning devices for the phantom and clinical studies, and evaluation of X-CARE with an anthropomorphic phantom) using Eq. (2) from Sec. 3.A. Additionally, this correction factor will be applied to measurements made as part of the upcoming clinical trial. Previous studies have shown a depletion effects that range from less than 0.1% per readout to 1.8% per readout,4,7,10,12 compared to the observed depletion rate of approximately 0.3%. This variation may be due, in part, to differences in the readout devices used in these studies,13,15 since one of these studies used a custom-built readout device.10 Another study characterized effects for much higher energy (6 MeV) monoenergetic photons beams.7 Of the studies that used the commercially available Microstar reader with nanoDots exposed to diagnostic energies, depletion effects only ranged between 0.27% (Ref. 12) and 0.5%.4 This is in very good agreement with the current results but emphasizes the importance in considering exposure and readout conditions in evaluating depletion effects. Other studies looking at the use of nanoDots for CT measurements did not include an evaluation of depletion effects.5,9 Unlike the depletion experiment, the fading experiment resulted in data that were not well fit with a simple model. A relative drop in counts of approximately 2.3% was observed to occur in the first 2 h followed by relatively stable Medical Physics, Vol. 42, No. 4, April 2015

4.B. Angular dependence and dosimeter positioning

As indicated in Sec. 3.B, no significant angular dependence was observed for the scan conditions used. Consequently, small variations in the positioning of the dosimeters on the outer surface of the safety glasses are not expected to significantly affect the measured dose. Additionally, since the use of glasses had no statistically significant effect on dose measurements, the use of plastic glasses as a positioning device in clinical studies is well-justified. Furthermore, radiologist review of phantom images that included the glasses demonstrated that the glasses had no adverse effect on image quality. As discussed in Sec. 2.C, previous studies performed at diagnostic energies have shown no significant angular dependence. Unlike previous studies, the geometry used here provided radially asymmetric doses to dosimeters placed in the 12 o’clock phantom position. While it is true that helical scans may result in radially asymmetric doses to the dosimeters, especially for pitches greater than one, all scan protocols used in this study were either axial or performed with a pitch less than one. Based on this, it was decided that testing angular dependence under helical scanning conditions was unnecessary. 4.C. Conversion from nanoDot counts to absorbed dose

Section 3.C demonstrates that the conversion from counts to absorbed dose to tissue was dependent on the position in which they were placed in the phantom. This result seems reasonable, given that OSL dosimeters are energy-dependent at diagnostic energies16 and that the scatter conditions for the two locations in the phantom are different.17 However, as Fig. 5 indicates, at the low doses likely to be observed from face and orbits exams, the difference between the regression lines for the center and 12 o’clock data is inconsequential. In evaluating dosimetry for head protocols, this effect should be considered. It should be reiterated that the CTDI phantom was used because it allowed the 0.6 cm3 ion chamber to be positioned accurately and reproducibly, which was critical for the characterization studies. The authors were not confident that the ion chamber could be affixed to the glasses in a reproducible manner. Given the results for the center and 12 o’clock phantom positions, it is unlikely that there would be a significant difference in conversion factors if the dosimeters

1737

R. M. Marsh and M. Silosky: Organ-based mA modulation for face and orbits scans

had been positioned on the glasses at the doses commonly seen during face and orbits exams. Yusuf et al. found a calibration of 52.83 counts/mrad (1.9 × 10−4 mGy/count) when using a portable x-ray unit for calibration,12 a value that is substantially different from the slopes identified in this study. However, this conversion factor, like others reported in the literature, was calculated for a single dose at a particular beam quality.5 It is difficult to make a direct comparison to other studies that only looked at this relationship relative to the manufacturer’s calibrations performed at 80 kVp.16 In other studies, conversion factors were determined over a length of 10 cm for CTDI100 measurements9,10 and may not be applicable for point measurements. Zhang et al. used a different approach and developed organ-specific calibration factors for body dosimetry measurements based on Monte Carlo-derived estimates of HVL.18 This method requires Monte Carlo simulations that may not be readily available to many clinical researchers. Akyakin described a calibration technique for skin dose measurements in dental cone beam CT, but these calibrations were performed only at scanner isocenter in air.19 These studies also did not examine the effects of depletion or fading. 4.D. Clinical study design and preliminary phantom results

The anthropomorphic phantom study indicated that the use of X-CARE would likely reduce dose to the lens of the eyes by approximately 18% for CT scans of the face and orbits when the head is positioned with the orbits at isocenter. These data will be used to perform the power analysis required to determine the number of subjects necessary for a statistically robust study. However, it should be noted that for both the face and orbits and head protocols, a substantial dose reduction was observed simply by raising the phantom so that the head, rather than the orbits, was centered in the scanner. There is no obvious trend demonstrated by the CTDI phantom study (Fig. 6) and the reason for this is unclear. However, the data obtained using the CTDI phantom agreed with the results from the anthropomorphic phantom study (Fig. 7) for the table position to be used in clinical studies. When the phantoms were at the lowest positions in the gantry (orbits centered), doses were reduced by 21.1% and 19.7% for the head and face and orbits protocols using the CTDI phantom and 19% and 18% for the head and face and orbits protocols using the anthropomorphic phantom. This dose reduction is less than what has been seen in other studies20,21 and is likely due to differences in the position of the orbits relative to scanner isocenter. Based on these results, it may be worth reconsidering the appropriate clinical practice regarding patient positioning. While centering the anatomy of interest can theoretically optimize spatial resolution, it remains to be determined if the relatively small changes in positioning used in this study will have a significant effect on image quality.22 Additionally, the patient positioning within the gantry may affect traditional tube current modulation algorithms, a factor that is not addressed here.23 It should be noted that others have observed dose Medical Physics, Vol. 42, No. 4, April 2015

1737

reduction of approximately 30% for head scans while using XCARE.2,20,21 This value agrees well with the 29% observed in this study for head scans performed with the phantom centered in the gantry. 5. CONCLUSION This work consists of three major components, the first of which is a thorough characterization of the nanoDot OSL dosimeters using a methodology that can be generally implemented for clinical dosimetry studies. The effects of depletion and fading were evaluated and a conversion from counts to absorbed dose, appropriate for the clinical conditions to be tested, was identified. The next component involved the validation of a dosimeter positioning method to be used in clinical studies and an evaluation of the angular dependence of nanoDot readings for CT scans. Based on the observed results, the use of plastic safety glasses as a positioning device provides an effective, reproducible, and sanitary means of incorporating the dosimeters in patient scans. The final component of the study included a preliminary evaluation of X-CARE as a dose reduction technique for CT scans of the face and orbits. The results of each of these components will be used to design, conduct, and analyze data from an upcoming clinical study intended to evaluate the use of X-CARE in patient scans of the face and orbits.

a)Author

to whom correspondence should be addressed. Electronic mail: [email protected]; Telephone: 303-724-3760; Fax: 303-7243795. 1“Guide to Low Dose” Siemens HealthCare, https://www.healthcare.siem ens.com/siemens_hwem-hwem_ssxa_websites-context-root/wcm/idc/grou ps/public/@global/@imaging/documents/download/mdax/nju4/ edisp/guid e-to-low-dose-2013-00849104.pdf (2013), Accessed 8 Sep. 2014. 2X. Duan, J. Wang, J. Christner, S. Leng, K. Grant, and C. McCollough, “Dose reduction to anterior surfaces with organ-based tube-current modulation: Evaluation of performance in a phantom study,” AJR, Am. J. Roentgenol. 197, 689–695 (2011). 3M. Lungren, T. Yoshizumi, S. Brady, G. Toncheva, C. Anderson-Evans, C. Lowry, X. Zhou, D. Frush, and L. Hurwitz, “Radiation dose estimations to the thorax using organ-based dose modulation,” AJR, Am. J. Roentgenol. 199, W65–W73 (2012). 4R. M. Al-Senan and M. R. Hatab, “Characteristics of an OSLD in the diagnostic energy range,” Med. Phys. 38, 4396–4405 (2011). 5L. Lavoie, M. Ghita, L. Brateman, and M. Arreola, “Characterization of a commercially-available, optically-stimulated luminescent dosimetry system for use in computed tomography,” Health Phys. 101, 299–310 (2011). 6“nanoDot: Single point radiation assessments” Landauer Europe, http:// www.landauer.com/uploadedFiles/Resource_Center/microStar%20Calibra tion%20and%20Usage%20Instructions%2010-Jun-09.pdf (2011), Accessed 16 Sep. 2014. 7P. Jursinic, “Characterization of optically stimulated luminescent dosimeters, OSLDs, for clinical dosimetric measurements,” Med. Phys. 34, 4594–4604 (2007). 8C. Ruan, E. G. Yukihara, W. J. Clouse, P. B. R. Gasparian, and S. Ahmad, “Determination of multislice computed tomography dose index (CTDI) using optically stimulated luminescence technology,” Med. Phys. 37, 3560–3568 (2010). 9T. J. Vrieze, G. M. Sturchio, and C. H. McCollough, “Technical note: Precision and accuracy of a commercially available CT optically stimulated luminescent dosimetry system for the measurement of CT dose index,” Med. Phys. 39, 6580–6584 (2012). 10E. G. Yukihara, C. Ruan, P. B. R. Gasparian, W. J. Clouse, C. Kalavagunta, and S. Ahmad, “An optically stimulated luminescence system to

1738

R. M. Marsh and M. Silosky: Organ-based mA modulation for face and orbits scans

measure dose profiles in x-ray computed tomography,” Phys. Med. Biol. 54, 6337–6352 (2009). 11L. Dunn, J. Lye, J. Kenny, J. Lehmann, I. Williams, and T. Kron, “Commissioning of optically stimulated luminescence dosimeters for use in radiotherapy,” Radiat. Meas. 51-52, 31–39 (2013). 12M. Yusuf, A. Saoudi, N. Alothmany, D. Alothmany, S. Natto, H. Natto, N. I. Molla, N. Mail, A. Hussain, and A. A. Kinsara, “Characterization of the optically stimulated luminescence nanoDot for CT dosimetry,” Life Sci. J. 11, 445–450 (2014). 13M. S. Akselrod, L. Botter-Jensen, and S. W. S. McKeever, “Optically stimulated luminescence and its use in medical dosimetry,” Radiat. Meas. 41, S78–S99 (2007). 14American Association of Physicists in Medicine, “The measurement, reporting, and management of radiation dose in CT,” AAPM Report No. 96 (AAPM, College Park, MD, 2008), see www.aapm.org. 15P. B. R. Gasparian, F. Vanhavere, and E. G. Yukihara, “Evaluating the influence of experimental conditions on the photon energy response of Al2O3:C optically stimulated luminescence detectors,” Radiat. Meas. 47, 243–249 (2012). 16C. Yahnke, “Calibrating the microStar,” Landauer White Paper ( http://www. landauer.com/uploadedFiles/Resource_Center/microStar%20Calibration% 20and%20Usage%20Instructions%2010-Jun-09.pdf, Accessed 8 Sep. 2014).

Medical Physics, Vol. 42, No. 4, April 2015

1738

17J. Boone, “Dose spread functions in computed tomography: A Monte Carlo

study,” Med. Phys. 36, 4547–4554 (2009). Zhang, X. Li, Y. Gao, X. G. Xu, and B. Liu, “A method to acquire CT organ dose map using OSL dosimeters and ATOM anthropomorphic phantoms,” Med. Phys. 40, 081918 (9pp.) (2013). 19S. Akyalcin, J. D. English, K. M. Abramovitch, and X. J. Rong, “Measurement of skin dose from cone-beam computed tomography imaging,” Head Face Med. 9, 28–34 (2013). 20C. Yamauchi-Kawaura, M. Yamauchi, K. Imai, M. Ikeda, and T. Aoyama, “Image quality and age-specific dose estimation in head and chest CT examinations with organ-based tube-current modulation,” Radiat. Prot. Dosim. 157, 193–205 (2013). 21A. J. Reimann, C. Davison, T. Bjarnason, Y. Thakur, K. Kryzmyk, J. Mayo, and S. Nicolaou, “Organ-based computed tomography (CT) radiation dose reduction to the lenses: Impact on image quality for CT of the head,” J. Comput. Assisted Tomogr. 36, 334–338 (2012). 22T. Hara, K. Ichikawa, S. Sanada, and Y. Ida, “Image quality dependence on in-plane positions and directions for MDCT images,” Eur. J. Radiol. 75, 114–121 (2009). 23J. Gudjonsdottir, J. R. Svensson, S. Campling, P. C. Brennan, and B. Jonsdottir, “Efficient use of automatic exposure control systems in computed tomography requires correct patient positioning,” Acta Radiol. 50, 1035–1041 (2009). 18D.