TY - JOUR
AU - Furui, Shigeru
AB - Abstract To minimise the radiation dermatitis related to interventional radiology (IR), rapid and accurate dose estimation has been sought for all procedures. We propose a technique for estimating the patient skin dose rapidly and accurately using Monte Carlo (MC) simulation with a graphical processing unit (GPU, GTX 1080; Nvidia Corp.). The skin dose distribution is simulated based on an individual patient’s computed tomography (CT) dataset for fluoroscopic conditions after the CT dataset has been segmented into air, water and bone based on pixel values. The skin is assumed to be one layer at the outer surface of the body. Fluoroscopic conditions are obtained from a log file of a fluoroscopic examination. Estimating the absorbed skin dose distribution requires calibration of the dose simulated by our system. For this purpose, a linear function was used to approximate the relation between the simulated dose and the measured dose using radiophotoluminescence (RPL) glass dosimeters in a water-equivalent phantom. Differences of maximum skin dose between our system and the Particle and Heavy Ion Transport code System (PHITS) were as high as 6.1%. The relative statistical error (2 σ) for the simulated dose obtained using our system was ≤3.5%. Using a GPU, the simulation on the chest CT dataset aiming at the heart was within 3.49 s on average: the GPU is 122 times faster than a CPU (Core i7–7700K; Intel Corp.). Our system (using the GPU, the log file, and the CT dataset) estimated the skin dose more rapidly and more accurately than conventional methods. INTRODUCTION Interventional radiology (IR) produces minimally invasive surgery for patients, but the number of cases accompanied by radiation dermatitis has increased recently. Generally, IR operations require a long examination time. Furthermore, complex procedures have increased in number. High doses have led to increased radiation injury to patients and staff members [1–3]. Many reports in the relevant literature claim that the patient exposure dose depends on the type of procedure [3–5]. Therefore, monitoring the exposure dose as accurately as possible, preferably in real time, is extremely important to avoid injury from radiation. Nevertheless, conventional methods of dose evaluation in IR are difficult. Moreover, they provide uncertain results. For example, radiophotoluminescence (RPL) glass dosimetry is well known for its excellent precision and good properties such as stability (negligible fading) and uniformity of products [6]. The use of these dosimeters, however, requires great effort that entails manual handling, annealing, preheating, and readout. Particularly, annealing takes 1 h at 400°C. Preheating takes 1 h at 70°C. A dose area product (DAP) meter can measure DAP, which is a product of the irradiated dose and the area of the exposure fields. The International Electrotechnical Commission (IEC) specifies 25% tolerance for this device [7]. This uncertainty value is insufficient for dose control. In addition to this shortcoming, and unfortunately by its definition, DAP loses information related to exposed positions. Recently, several real-time dosimeters have become commercially available [8–12]. However, they measure doses only at one or at a few points on a patient. The Monte Carlo (MC) method is well known for use as a dose calculation algorithm. Nevertheless, it has also been notorious for its requirement of long durations to produce highly precise estimates. Recently, one means of reducing the calculation time was proposed by Badal et al. The method uses graphical processing units (GPUs). Actually, MC-GPU v1.3 [13] is a Monte Carlo simulation package for use with GPUs to generate synthetic radiographic images in the public domain. A few real-time skin dose estimation methods for IR have been reported [14, 15]. They estimate skin dose on one standard human model using a theoretical mathematical model. No report of the relevant literature has described a study using an individual patient’s body along with the MC method for estimating the detailed skin dose distribution for IR. This paper presents a proposal of a novel fast dose estimation system for IR (FDEIR) using an individual patient’s computed tomography (CT) dataset. Fluoroscopy conditions extracted from examination records were fed to the simulation system automatically. In this way, the patient skin dose is calculable immediately after fluoroscopic examination, with accuracy of 6.1%. MATERIALS AND METHODS MC-GPU–based system FDEIR is based on MC-GPU, a public domain MC simulation package that can simulate synthetic radiographs and computed tomography (CT) scans using the computational power of commercially available GPU cards [13]. The package uses the PENELOPE 2006 subroutine [16] for photon interaction models, photon transport models, and material cross sections. The Woodcock tracking algorithm [17] for photon transport and suppression of electron transport is also implemented to accelerate calculation in the package. For calculations, the package requires the x-ray source position, the source trajectory for the CT scan, the x-ray energy spectrum, a random seed, the number of simulated histories and GPU that threads use, a voxelized geometry definition, the detector structure, and material files. FDEIR framework We applied MC-GPU for absorbed dose evaluation of a patient’s skin, particularly addressing IR dose evaluation, using an individual CT dataset and built FDEIR. Repeated simulations were conducted for FDEIR on a voxelized CT phantom with different conditions, extracted from the log file of a fluoroscopic examination. The absorbed skin dose was estimated by converting the simulated relative dose to an absolute dose. In FDEIR, the x-ray source is defined as a point source. We omitted some unnecessary parts from the package for synthetic radiographic image generation. For MC simulation, FDEIR requires fluoroscopic conditions that include tube voltage, milliampere–second (mAs) values, source-to-image-receptor distance (SID), field of view (FOV) size, source position, and imaging table position, which are all included in a fluoroscopic log file. For every exposure, FDEIR reads them out. The energy spectra are indispensable for MC simulation. They were computed in an earlier study using spectrum computation software with the Birch and Marshall model [18, 19]. In FDEIR, a CT phantom is constructed from the patient CT dataset before dose calculation. Then, FDEIR reads in the phantom and material data, along with the calendar time (seconds since 1 Jan 1970) (as a random seed to prevent production of the same result) and the fluoroscopic condition from the log file in random access memory (RAM). Subsequently, these data are copied to the GPU. Next, FDEIR executes the MC simulation. Outputs simulate the skin dose distribution. After simulation with each fluoroscopic condition, the simulated dose distribution is converted into the absorbed dose distribution using a linear regression equation, as described in the Computing scaling factor section herein. Similarly to the original MC-GPU, electron transport is suppressed in FDEIR for accelerating the calculation, because the electron attenuation length at the diagnostic energy range in water is <300 μm [20]. Moreover, it is much shorter than the minimum voxel size of the CT dataset (1.074 × 1.074 × 5 mm). Hardware specifications used for this study included a CPU (Core i7–7700K; Intel Crop.), 64 GB RAM and a GPU (GTX 1080; Nvidia Corp.). System validation To validate FDEIR, a comparison with Monte Carlo code Particle and Heavy Ion Transport code System (PHITS) ver. 2.88 [21] and the measurements was made. Electron Gamma Shower Version 5 (EGS5) [22] is incorporated for photon and electron transport in PHITS. Their transport in EGS5 was confirmed [23, 24]. This validation ensured that FDEIR has a plausible transport model and no need for electron transport. For comparison, the percentage depth dose (PDD) curves for a 76 kV x-ray beam were evaluated in water. The electron transport was simulated in PHITS. The cut-off energy was 5 keV in each case. An RPL glass dosimeter (GD-352M; Chiyoda Technol Corp.) was used for measurements. An annealing process was applied before exposing the RPL glass dosimeter to x-rays. The dose was measured every 2.5 cm along the x-ray beam axis (Fig. 1). Fig. 1. View largeDownload slide Geometry arrangement for measurements and simulations to evaluate dose in the water-equivalent phantom. In the simulation, the imaging table is assumed to be a graphite board. Glass dosimeters are sandwiched between MIX-R and the imaging table. Source-to-surface distances (SSDs) were 60 cm and 50–70 cm, respectively, for PDD and the scaling factor. Fig. 1. View largeDownload slide Geometry arrangement for measurements and simulations to evaluate dose in the water-equivalent phantom. In the simulation, the imaging table is assumed to be a graphite board. Glass dosimeters are sandwiched between MIX-R and the imaging table. Source-to-surface distances (SSDs) were 60 cm and 50–70 cm, respectively, for PDD and the scaling factor. Additionally, we compared the dose at the center of the surface of the water-equivalent phantom as determined by FDEIR, PHITS and measurement. We used the same geometry presented in Fig. 1. Similarly, we compared the relative skin dose distribution between FDEIR and PHITS. PHITS constructs a voxelized phantom using the DICOM2PHITS program, which is a part of PHITS for inputting the CT DICOM dataset [21, 25]. Actually, FDEIR constructs it using a technique described in the Voxelization from a CT dataset section herein. For this comparison, we assumed for cardiac fluoroscopy that the chest was exposed to the x-ray beam in the posterior–anterior (PA) direction (Fig. 2). Fig. 2. View largeDownload slide Geometry arrangement for skin dose simulation in FDEIR and PHITS. The phantom size is 43.0 × 25.8 × 36.0 cm3. The beam was incident on the center of the phantom, aimed at the heart. Fig. 2. View largeDownload slide Geometry arrangement for skin dose simulation in FDEIR and PHITS. The phantom size is 43.0 × 25.8 × 36.0 cm3. The beam was incident on the center of the phantom, aimed at the heart. Computing scaling factor The original MC-GPU returns the absorbed dose divided by the number of incident photons. Determination of the number of incident photons from a fluoroscopic apparatus makes it possible to estimate the x-ray beam dose that is absorbed. We therefore defined the number as a scaling factor. The number of incident photons depends on the energy spectrum and the irradiation field size. Therefore, the scaling factor must be defined for every energy spectrum and irradiation field size. For this study, we adopted a 76 kV spectrum, 20 cm FOV and 100 cm SID. The number of incident photons is proportional to the mAs value. The scaling factor was also associated with the mAs value. The measured absorbed dose was proportional to the simulated relative dose at a reference point. The reference point is the center of a boundary between a water-equivalent phantom and an imaging table. The geometry is presented in Fig. 1. The absolute doses at the reference point were measured using a RPL glass dosimeter (GD-352M) with application of the annealing process. The dose figures represent the average of two RPL glass dosimeters placed at the reference point. For the simulation, a voxelized geometry was constructed with similar geometry of the measurement (Fig. 1). The fluoroscopy conditions were extracted from the log file. Then, the relation between the simulated dose and that measured using RPL glass dosimeters in a water-equivalent phantom was approximated using a linear function with a zero intercept. The linear regression equation between them was y = 0.203x. Voxelization from a CT dataset For this study, a voxelized phantom was constructed from an anonymized chest CT dataset (1.074 × 1.074 × 5 mm, Fig. 3) provided by Teikyo University Hospital. To reduce statistical error and loading time of the phantom, the CT dataset was downsampled to 5.37 × 1.074 × 5 mm by taking the average of pixel values. Fig. 3. View largeDownload slide Volume rendered 3D chest CT dataset (left) and a slice of the voxelized phantom at the heart level (right). The voxelized phantom is segmented into five regions: air, water, bone, skin and graphite. Fig. 3. View largeDownload slide Volume rendered 3D chest CT dataset (left) and a slice of the voxelized phantom at the heart level (right). The voxelized phantom is segmented into five regions: air, water, bone, skin and graphite. Air, water and bone were distinguished within the CT dataset, based on pixel values, to allocate the tissue type for each pixel. The values of air, water and bone are mutually distinct, which simplifies their categorization. In contrast, categorizing organs and soft tissues is difficult because their values are very similar. Organs and soft tissues have values that are approximately equal to that of water. We treated organs and soft tissues, except for skin, as water for construction of the voxelized phantom. The phantom skin was assumed to be one voxel layer at the outer surface of the body. The voxel thickness is 1.074 mm, which is approximately equal to the reference value of the ICRP: 1.3 mm [26]. Material data for air, water, bone and skin were obtained from the MC-GPU package. Attenuation and scattering at an imaging table are important factors for calculating the skin dose. Therefore, we added a graphite board (3.8 cm Innova 4100-IQ; GE Healthcare) to the phantom as the imaging table at Teikyo University Hospital. Material data were based on the database from PENELOPE. A slice of the voxelized phantom is portrayed in Fig. 3. Rapid dose calculation At the start of each simulation, the MC-GPU reads the voxelized phantom and material data from a hard drive to memory. When using the chest CT phantom, the reading time is ~1 s. It is a large fraction of GPU simulation time. FDEIR reads these data only once at the beginning of the process. To measure the calculation time, we simulated 10 million and 100 million histories with Case A (Table 1) in FDEIR on the computer. Similarly, these simulations were executed in PHITS. Table 1. Fluoroscopic conditions of X-ray beams aimed at the heart Case  SSD  kV  FOV  SID  Direction  A  50  76  20  100  PA  B  60  76  20  100  PA  C  70  76  20  100  PA  D  60  76  20  100  RAO30  Case  SSD  kV  FOV  SID  Direction  A  50  76  20  100  PA  B  60  76  20  100  PA  C  70  76  20  100  PA  D  60  76  20  100  RAO30  We compared the maximum skin doses estimated using FDEIR, PHITS, and the dose recorded in the log file. The log file shows values calculated using a commercial apparatus (Innova 4100-IQ; GE Healthcare) at Teikyo University Hospital. The values also include maximum absorbed dose records in mGy units for each exposure. However, the value was rounded to the nearest integer. The recorded doses were extracted from the log file with each fluoroscopy condition. In FDEIR and PHITS, the maximum skin dose was estimated using these conditions. The conditions are presented in Table 1. The geometry is depicted in Fig. 2. RESULTS System validation Validation of FDEIR was performed using a comparison with PHITS. Figure 4 shows PDD curves evaluated using FDEIR, PHITS, and measurement in the water phantom. The difference between them, i.e. [(our FDEIR—PHITS dose)/PHITS dose], was 4.3% on average. That result implies that FDEIR has plausible transport of photons, with no necessity of electron transport in the diagnostic energy range. Fig. 4. View largeDownload slide PDD curves evaluated using FDEIR, PHITS, and measurement in a water phantom for 76 kV x-ray beam at 60 cm SSD. Error bars show variations in the reading reproducibility (5%). Fig. 4. View largeDownload slide PDD curves evaluated using FDEIR, PHITS, and measurement in a water phantom for 76 kV x-ray beam at 60 cm SSD. Error bars show variations in the reading reproducibility (5%). Compute absorbed dose Figure 5 presents measured and simulated doses with the least-squares fitting straight line for computing the scaling factor. The linear regression equation between them was y = 0.203x. The coefficient of determination, the square of Pearson’s correlation coefficient (r = 0.998), was 0.995. That result shows that measured doses and simulated ones are proportional and that linear regression is plausible. For this study, we defined the linear regression equation as the scaling factor. Fig. 5. View largeDownload slide The scaling factor to convert from relative dose to absolute dose is the slope of the line. Points on this graph correspond to the measurements and simulation dose on the geometry shown in Fig. 1. Error bars show variations in the reading reproducibility (5%). Fig. 5. View largeDownload slide The scaling factor to convert from relative dose to absolute dose is the slope of the line. Points on this graph correspond to the measurements and simulation dose on the geometry shown in Fig. 1. Error bars show variations in the reading reproducibility (5%). We provide a couple of examples of the absolute skin dose distribution in Figs 6 and 7. They show the absorbed skin dose distribution for the part of the skin in the chest CT phantom. The maximum dose was estimated as 0.573 mGy/mAs. Figure 7 portrays the skin dose distribution in the chest CT phantom, calculated respectively using FDEIR and PHITS. Those two distributions are apparently consistent. The error in the irradiation field was 5.8% on average. The difference derives from statistical error or the construction methods for the phantom, which was made using image thresholding in FDEIR; in PHITS, the phantom was constructed using DICOM2PHITS. Fig. 6. View largeDownload slide Absorbed skin dose distribution on the front (left) and the back (right) of the chest in the chest CT phantom. There were 100 million incident photons. The x-ray beam used for fluoroscopy of the heart had the following specifications: 76 kV, 20 cm FOV, 100 cm SID, and 60 cm SSD in the PA direction. Fig. 6. View largeDownload slide Absorbed skin dose distribution on the front (left) and the back (right) of the chest in the chest CT phantom. There were 100 million incident photons. The x-ray beam used for fluoroscopy of the heart had the following specifications: 76 kV, 20 cm FOV, 100 cm SID, and 60 cm SSD in the PA direction. Fig. 7. View largeDownload slide Absorbed skin dose distribution in 2D for an area of the skin on the back of the thorax of the CT phantom (left) and axial dose distributions indicated in a line shown in the left panel (right). The PHITS dose was normalized to the simulated maximum dose from FDEIR. The X mark shows the maximum dose position. It is 0.573 mGy/mAs. Fig. 7. View largeDownload slide Absorbed skin dose distribution in 2D for an area of the skin on the back of the thorax of the CT phantom (left) and axial dose distributions indicated in a line shown in the left panel (right). The PHITS dose was normalized to the simulated maximum dose from FDEIR. The X mark shows the maximum dose position. It is 0.573 mGy/mAs. Comparison of the absorbed doses The absorbed doses at the center of the surface of the water-equivalent phantom were evaluated. The measured dose was 3.66 ± 0.18 mGy when the phantom was exposed to the x-ray beam for 30 s with Case B (Table 1). The dose was calculated with FDEIR as 3.71 mGy under the same conditions. This result and the others are presented in Fig. 8. The difference between the absorbed dose from FDEIR and that from measurement was 2.9% on average. Fig. 8. View largeDownload slide Comparison of absorbed dose in the water-equivalent phantom between measurements, total dose on the log file, simulated absorbed dose by FDEIR, and PHITS dose, where the PHITS dose was normalized by the simulated absorbed dose at SSD60. Fig. 8. View largeDownload slide Comparison of absorbed dose in the water-equivalent phantom between measurements, total dose on the log file, simulated absorbed dose by FDEIR, and PHITS dose, where the PHITS dose was normalized by the simulated absorbed dose at SSD60. Regarding the skin dose, the log file recorded the absorbed dose as 3 mGy and the mAs value as 0.16234 per second when the phantom was exposed to the x-ray beam for 30 s with Case B (Table 1). We calculated the absorbed dose using FDEIR under the same conditions. The simulated absorbed dose was 2.91 mGy in the voxel exposed maximum dose. When the x-ray beam was aimed in the RAO 30 direction (Table 1), the recorded dose was 6 mGy. The simulated absorbed dose was 5.22 mGy. Figure 9 presents these results and other results for some conditions, where the PHITS dose was normalized by the simulated absorbed dose at SSD60. The relative statistical error (2 σ) was 3.5% at the maximum. Differences between the maximum skin dose between FDEIR and PHITS were as great as 6.1%. Fig. 9. View largeDownload slide Comparison of absorbed dose of the skin in the chest CT phantom between total dose on the log file, simulated absorbed dose by FDEIR, and PHITS dose, where the PHITS dose was normalized by the simulated absorbed dose at SSD60. Fig. 9. View largeDownload slide Comparison of absorbed dose of the skin in the chest CT phantom between total dose on the log file, simulated absorbed dose by FDEIR, and PHITS dose, where the PHITS dose was normalized by the simulated absorbed dose at SSD60. Calculation time Simulation of 10 million and 100 million incident particles on the chest CT phantom with Case A (Table 1) was executed on the computer. Table 2 presents a summary of the calculation times of the simulations. For the simulation of 100 million incident photons, the GPU (GTX 1080) was ~122 times as fast as the single-threaded CPU (Core i7–7700K). Table 2. Simulation execution times on the voxelized chest CT phantom Simulated photons  GTX 1080  i7–7700K  PHITS  1.00E+07  1.15 s  49 s  100 min  1.00E+08  3.49 s  427 s  954 min  Simulated photons  GTX 1080  i7–7700K  PHITS  1.00E+07  1.15 s  49 s  100 min  1.00E+08  3.49 s  427 s  954 min  DISCUSSION The simulation results for maximum skin dose showed agreement between FDEIR and the conventional method (Figs 7 and 9). The dose from the log file was rounded to the nearest integer. The electron transport was simulated in PHITS. Results show two salient benefits of FDEIR. First, the exposed location on the skin is identifiable in the calculated 3D-dose grid. Although an exposed position is indispensable for estimation of the skin dose distribution, the log file shows not the dose distribution, but the maximum absorbed dose, excluding exposed positions. Accurate knowledge related to the skin dose distribution facilitates prediction of the extent of radiation dermatitis. Second, the simulation can be performed rapidly using the GPU. Estimation of the skin dose distribution can be done immediately after an examination. Moreover, with extraction of fluoroscopic examinations from the log file during an examination, FDEIR is useful for real-time estimation. Cognition of the exposed position and dose help to suppress high-dose exposure at any one specific point. The occurrence of radiation dermatitis can therefore be controlled. Comparison of the simulation times for 10 million incident particles and 100 million incident particles on the same hardware revealed that the GPU (GTX 1080) took about three times longer, although the incident particles were 10 times more numerous. Results suggest that reading the patient CT phantom and material data from RAM to GPU global memory accounts for a large fraction of the execution time. For further acceleration of the simulation, these data should be retained in GPU global memory during simulations. For this study, FDEIR was used for skin dose estimation. It provides a detailed dose distribution for internal organs if the patient CT phantom is constructed elaborately, with inclusion of internal organs. Furthermore, by altering the patient CT phantom to an IR room phantom, the air dose distribution in the IR room can be estimated rapidly. This makes it easier to identify points with a low air dose rate. In conclusion, this paper presents FDEIR based on the MC-GPU package. Use of individual CT datasets and various fluoroscopic conditions strongly affects accurate dose estimation. Simulation using the GPU supports rapid calculation. FDEIR is expected to be helpful for ensuring radiation safety for IR. CONFLICT OF INTEREST The authors declare that there are no conflicts of interest. FUNDING This work was partly supported by a Japan Society for the Promotion of Science (JSPS) KAKENHI Grant [Number 15K08703]. REFERENCES 1 Sánchez RM, Vano E, Fernández JM et al.  . Staff doses in interventional radiology: a national survey. J Vasc Interv Radiol  2012; 23: 1496– 501. Google Scholar CrossRef Search ADS PubMed  2 Food and Drug Administration. Public Health Advisory: avoidance of serious x-ray-induced skin injuries to patients during fluoroscopically-guided procedures. https://www.fda.gov/downloads/Radiation-EmittingProducts/RadiationEmittingProductsandProcedures/MedicalImaging/MedicalX-Rays/ucm116677.pdf (1 June 2017, date last accessed). 3 Pantos I, Patatoukas G, Katritsis DG et al.  . Patient radiation doses in interventional cardiology procedures. Curr Cardiol Rev  2009; 5: 1– 11. Google Scholar CrossRef Search ADS PubMed  4 Zorzetto M, Bernardi G, Morocutti G et al.  . Radiation exposure to patients and operators during diagnostic catheterization and coronary angioplasty. Cathet Cardiovasc Diagn  1997; 40: 348– 51. Google Scholar CrossRef Search ADS PubMed  5 Neofotistou V, Vano E, Padovani R et al.  . Preliminary reference levels in interventional cardiology. Eur Radiol  2003; 13: 2259– 63. Google Scholar CrossRef Search ADS PubMed  6 Murakami H, Takahashi F, Griffith RV. A Personal Dosimetry Intercomparison Study in Asian and Pacific Region. http://www.irpa.net/irpa10/cdrom/00184.pdf (1 June 2017, date last accessed). 7 IEC 60580 Ed. 2.0: Medical Electrical Equipment—Dose Area Product Meters. 2000. 8 Nakamura M, Chida K, Zuguchi M. Novel dosimeter using a nontoxic phosphor for real-time monitoring of patient radiation dose in interventional radiology. AJR Am J Roentgenol  2015; 205: W202– 6. Google Scholar CrossRef Search ADS PubMed  9 Ishikawa M, Nagase N, Matsuura T et al.  . Development of a wavelength-separated type scintillator with optical fiber (SOF) dosimeter to compensate for the Cerenkov radiation effect. J Radiat Res  2015; 56: 372– 81. Google Scholar CrossRef Search ADS PubMed  10 Ishikawa M, Bengua G, Sutherland KL et al.  . A feasibility study of novel plastic scintillation dosimetry with pulse-counting mode. Phys Med Biol  2009; 54: 2079– 92. Google Scholar CrossRef Search ADS PubMed  11 Hwang E, Gaxiola E, Vlietstra RE et al.  . Real-time measurement of skin radiation during cardiac catheterization. Cathet Cardiovasc Diagn  1998; 43: 367– 70. Google Scholar CrossRef Search ADS PubMed  12 Nishikido F, Moritake T, Ito H et al.  . A prototype real-time dose distribution monitoring system using plastic scintillators connected to optical fiber for interventional radiology. 2013 IEEE Nucl Sci Symp Med Imaging Conf (2013 NSS/MIC) 2013:1–3. 13 Badal A, Badano A. Accelerating Monte Carlo simulations of photon transport in a voxelized geometry using a massively parallel graphics processing unit. Med Phys  2009; 36: 4878. Google Scholar CrossRef Search ADS PubMed  14 den Boer A, de Feijter PJ, Serruys PW et al.  . Real-time quantification and display of skin radiation during coronary angiography and intervention. Circulation  2001; 104: 1779– 84. Google Scholar CrossRef Search ADS PubMed  15 Cusma JT, Bell MR, Wondrow MA et al.  . Real-time measurement of radiation exposure to patients during diagnostic coronary angiography and percutaneous interventional procedures. J Am Coll Cardiol  1999; 33: 427– 35. Google Scholar CrossRef Search ADS PubMed  16 Salvat F, Fernández-Varea J, Sempau J. PENELOPE-2006: A Code System for Monte Carlo Simulation of Electron and Photon Transport. https://www.oecd-nea.org/science/pubs/2006/nea6222-penelope.pdf (1 June 2017, date last accessed). 17 Sempau J, Wilderman SJ, Bielajew AF. DPM, a fast, accurate Monte Carlo code optimized for photon and electron radiotherapy treatment planning dose calculations. Phys Med Biol  2000; 45: 2263– 91. Google Scholar CrossRef Search ADS PubMed  18 Kato H. Monte Carlo simulation by personal computers. Jpn J Radiol Technol  1992; 55: 190– 4 (in Japanese). Google Scholar CrossRef Search ADS   19 Birch R, Marshall M. Computation of bremsstrahlung X-ray spectra and comparison with spectra measured with a Ge(Li) detector. Phys Med Biol  1979; 24: 505– 17. Google Scholar CrossRef Search ADS PubMed  20 Berger MJ, Coursey JS. DSZ. ESTAR: Stopping Powers and Ranges for Electrons. National Institute of Standards and Technology (NIST). http://physics.nist.gov/PhysRefData/Star/Text/ESTAR.html (1 June 2017, date last accessed). 21 Sato T, Niita K, Matsuda N et al.  . Particle and heavy ion transport code system, PHITS, version 2.52. J Nucl Sci Technol  2013; 50: 913– 23. Google Scholar CrossRef Search ADS   22 Hirayama H, Namito Y, Bielajew AF et al.  . The EGS5 code system, SLAC-R-730 and KEK report 2005–08. Technical Report. Stanford Linear Accelerator Center and Tsukuba: High Energy Accelerator Research Organization, Stanford, 2005. 23 Mohammadi A, Baba M, Oishi T. Measurement and calculation of Compton Spectrometer response function for 10–60-keV monoenergetic photons. J Nucl Sci Technol  2010; 47: 570– 4. Google Scholar CrossRef Search ADS   24 Kajimoto T, Endo S, Tat Thanh N et al.  . Calculation of coincidence summing in gamma-ray spectrometry with the EGS5 code. Appl Radiat Isot  2015; 95: 53– 8. Google Scholar CrossRef Search ADS   25 Furuta T, Hashimoto S, Sato T. Medical applications of the PHITS Code (3): User Assistance Program for Medical Physics Computation. Jpn J Med Phys  2016; 36: 50– 4 (in Japanese). 26 Sato K, Noguchi H, Emoto Y et al.  . Japanese adult male voxel phantom constructed on the basis of CT images. Radiat Prot Dosimetry  2007; 123: 337– 44. Google Scholar CrossRef Search ADS PubMed  © The Author 2017. Published by Oxford University Press on behalf of The Japan Radiation Research Society and Japanese Society for Radiation Oncology. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com
TI - Fast skin dose estimation system for interventional radiology
JF - Journal of Radiation Research
DO - 10.1093/jrr/rrx062
DA - 2018-03-01
UR - https://www.deepdyve.com/lp/oxford-university-press/fast-skin-dose-estimation-system-for-interventional-radiology-eae0DiF9mt
SP - 233
EP - 239
VL - 59
IS - 2
DP - DeepDyve
ER -