Article ID: 25-00015
Abstract
We clarified the differences between the two pre-set protocols for iodine-digital subtraction angiography (ICM-DSA) and carbon dioxide (CO2)-DSA in terms of radiation exposure to patients and physicians when patients diagnosed with lower extremity artery disease (LEAD) undergo ICM-DSA or CO2-DSA, and identified factors that affect radiation exposure when patients undergo ICM-DSA or CO2-DSA. CO2-and ICM-DSA yielded 149 and 272 vessels in 21 and 49 patients, respectively. Multiple regression analyses revealed strong positive linear correlations between total air kerma (kinetic energy released per unit mass) at the patient entrance reference point (Ka,r), the total kerma area product (KAP), and the total number of acquisitions for ICM-DSA and CO2-DSA, respectively. Compared to Ka,r, there were significant differences in the tube potentials, KAP, and spectral shaping filters in both DSA procedures. The linear correlations between Ka,r and the number of ICM-DSA and CO2-DSA images varied among dose types, with strong linear correlations observed when Ka,r was classified into the above three groups. The X-ray conditions for each DSA image differed, with CO2-DSA using a significantly higher tube potential and different spectral shaping filters compared to ICM-DSA. When performing DSA for LEAD, an optimized protocol for each type of ontrast agent should be used.
Lower extremity arterial disease (LEAD) occurs when plaques accumulate in the leg arteries, creating a blockage. Catheter-based angiography using iodine contrast media (ICM) remains the gold standard for diagnosing LEAD but is now limited to patients receiving endovascular revascularization1). Carbon dioxide (CO2) DSA has been used to replace or supplement (to reduce contrast) conventional contrast-based angiography2). The use of ICM in patients with impaired renal function may lead to contrast-induced nephropathy (CIN)3). ICM is contraindicated in patients with an iodine allergy. Many patients benefit from interventional radiology (IR) procedures such as percutaneous coronary intervention;however, radiation exposure is a major disadvantage4-7). These procedures tend to be complex and require relatively lengthy fluoroscopy times8,9). Consequently, the radiation doses to both physicians and patients are high, and radiation protection is warranted10-16). Deterministic radiation effects are a significant concern, as the number of radiation-related skin injuries caused by IR procedures is increasing17-25).
One phantom study reported that ICM-DSA and CO2-DSA had different incident surface air kerma (kinetic energy released per unit mass) rates depending on the body thickness26). However, to the best of our knowledge, no studies have assessed the actual doses required to acquire ICM-DSA and CO2-DSA images in a LEAD setting, and the factors determining these doses have not been identified. This retrospective study aimed to explore the differences between two preset DSA protocols, ICM-DSA and CO2-DSA, in terms of radiation dose for patients and physicians, as well as the factors affecting radiation dose in patients diagnosed with LEAD undergoing ICM-DSA or CO2-DSA.
The study population consisted of adults aged ≥ 18 years who were diagnosed with LEAD. Only patients who underwent DSA using the Artis Zee ceiling system (Siemens Healthcare, Forchheim, Germany) were included to eliminate the bias caused by equipment differences. Continuous ICM-DSA or CO2-DSA data were accumulated from 1 April 2013 to 31 March 2019, when the physician acquired the ICM-DSA or CO2-DSA image. The selection of vessels for ICM-DSA and CO2-DSA, as well as the timing of acquisition, was determined by a single physician. The Digital Imaging and Communications in Medicine (DICOM) header of the DSA images were recorded as follows:patient background (sex, age, weight, and height);The following procedural parameters were recorded:tube potential, tube current, pulse width, spectral shaping filter, frame rate, source-to-flat panel detector [FPD] distance, source-to-patient distance, field of view, air kerma at patient entrance reference point [Ka,r], kerma area product [KAP], total number of acquisitions, total Ka,r, total KAP, total fluoroscopy time, and total number of images. The body mass index (BMI) was calculated using height and weight. The patient-to-FPD distance is calculated by subtracting the source-to-patient distance from the source-to-FPD distance. Normalized by the number of acquisitions, the dose per acquisition was obtained from the total Ka,r and total KAP. The total Ka,r is the air kerma at the patient entrance reference point accumulated over the entire X-ray procedure, and the total KAP is the integral of the air kerma in free air over the area of the X-ray beam in the plane perpendicular to the beam center axis for the entire X-ray procedure. Ka,r and KAP are the values for each DSA series acquired once, not for the entire procedure. Normalized Ka,r and normalized KAP are the total Ka,r and total KAP divided by the total number of acquisitions, respectively. Classification was performed based on the ratio of Ka,r to the number of DSA images, using the dose required to construct a single DSA image. The dose threshold was set at a value that resulted in a high correlation coefficient between Ka,r and the number of DSA images. Groups with Ka,r greater than the number of DSA images were classified as type A, groups with Ka,r smaller than the number of DSA images were classified as type C, and those in between were classified as type B.
Image acquisitionICM-DSA and CO2-DSA images were acquired using the Artis zee Ceiling system equipped with an FPD and a grid (grid ratio 15:1, grid density 80/cm, focusing distance 105 cm), which was assessed regularly to ensure optimal imaging performance and an adequate radiation dose27-32). A KAP meter (K1-S, PTW, Freiburg, Germany) was integrated into the angiography system. Images were acquired under automatic dose-rate control (ADRC), and the ICM-DSA and CO2-DSA protocols were selected by the radiological technologist responsible for the procedure. The ADRC automatically controls tube current, pulse width, and focal spot size to provide a consistent FPD input dose. The tube potential and FPD input dose rate for ICM-DSA were set at 70 kV and 2.400 μGy/fr, respectively;the corresponding settings for CO2-DSA were 81 kV and 3.600 μGy/fr, respectively. The FPD input dose rate for CO2-DSA was preset at 1.5 times that for ICM-DSA (2.400 vs 3.600 μGy/fr). Due to the low X-ray absorption of CO2, a high dose was required to obtain image quality equivalent to that of ICM-DSA. In addition, the frame rate for both protocols was set to 6 frames/s, and the maximum pulse width was set to 50 ms by the vendor. The frame rate can be adjusted by a radiological technologist, depending on the imaging site or the type of contrast media used. The spectral shaping filter was set from a minimum of 0 mmCu to a maximum of 0.6 mmCu. Furthermore, a recursive filter was used in either of the DSA protocols. Recursive filters are designed to reduce noise in DSA images.
Spatial scattered radiation dose rate evaluationIn a previous study, the patient doses of polymethyl methacrylate (PMMA) were assessed26). The fluoroscopy control curves for CO2-DSA were divided into four phases based on the thickness of the spectral-shaping filter. Similarly, the fluoroscopy control curves of ICM-DSA were divided into three phases. As a result, below 20 cm of PMMA, the radiation doses to patients and physicians in CO2-DSA were 20.7 ± 13.6% and 20.1 ± 12.0% lower, respectively, than those in ICM-DSA. Therefore, this study used PMMA with a thickness of 20 cm as a phantom. The measurement position was 100 cm from the caudal side of the image center and 100 cm from the operator side. The measurement position was assumed to be that of the physician near the surgical site. Measurements were taken at a height of 150 cm to evaluate the effect of the scattered radiation from the patient and the FPD on the physician. An ionization chamber survey meter (ICS-311;Aloka Co. Ltd., Tokyo, Japan) was used. ICM-DSA and five X-ray conditions of CO2-DSA (I, II, III, IV, V) were conducted (Table 1). The five X-ray conditions were set by the vendor at an input dose rate above and below 3.600 μGy/fr, which is the value set on the device. A protocol was created with an input dose rate of 5.400 μGy/fr and a tube potential set to 70 kV, the same as ICM-DSA, to find the effect of different tube potentials. The spatially scattered radiation dose rate measurement positions are shown in Figure 1. The number of vessels as well as the number of patients was recorded to calculate the percentage of arteries contrasted by ICM-DSA and CO2-DSA.
Statistical analysisThe height, weight, age, total Ka,r, total KAP, total fluoroscopy time, and total number of imaging procedures were compared between the ICM-DSA and CO2-DSA groups. Differences in means were analyzed using Welch’s t-test and the Wilcoxon rank-sum test. Statistical significance was set at p < 0.05. Additionally, X-ray conditions attributable to image acquisition were examined using multiple regression analysis. The correlation coefficient is expressed as r, and the coefficient of determination as R2. Furthermore, the slope of the approximate line is a. Statistical analyses were conducted using R v. 4.4.2 for the macOS X Cocoa GUI (https://www.r-project.org/)
X-ray conditions of ICM-DSA and CO2-DSA
DSA, digital subtraction angiography; ICM-DSA, iodine contrast medium digital subtraction angiography; CO2-DSA, carbon dioxide digital subtraction angiography; FPD, flat panel detector
Experimental setup for the dose rate measurements. Abbreviations:FPD, flat panel detector.
Angiography was performed from the abdomen to the feet in patients with LEAD. ICM-DSA yielded 272 vessels in 49 patients (20 women [40.8%];mean age, 75.3 ± 13.5 years), and CO2-DSA yielded 148 vessels in 21 patients (7 women [33.3%];mean age, 75.1 ± 11.1 years). The years of experience of the radiological technologists who participated in the IR procedure ranged from 1 to 40 years. Whether an equalization filter was used during the procedure was unknown in this retrospective study. However, they were not used in DSA. The arteries examined using DSA are shown in Table 2. Arteries were defined as those in which the catheter tip was placed. There was a correlation between ICM-DSA and CO2-DSA for the total Ka,r or total KAP. There were no significant differences in the total Ka,r, age, height, weight, BMI, total number of acquisitions, or total fluoroscopy time between the two groups (Table 3). The total Ka,r for ICM-DSA was 950.4 ± 1,130.2 mGy, and that for CO2-DSA was 1,093.1 ± 703.4 mGy (p = 0.50). The total KAP for ICM-DSA was 116.8056 ± 115.8128 Gy cm2, and that for CO2-DSA was 161.3775 ± 99.2985 Gy cm2 (p = 0.78). The normalized Ka,r results were 45.9 ± 43.3 mGy for ICM-DSA and 60.2 ± 41.5 mGy for CO2-DSA (Figure 2;p = 0.11), and the normalized KAP results were 6.326 ± 6.045 for ICM-DSA and 9.829 ± 9.246 Gy cm2 for CO2-DSA (Figure 3;p < 0.05). Total KAP (ICM-DSA vs CO2-DSA:1.4 times), normalized KAP (1.6 times), and normalized Ka,r (1.3 times) were close to the set FPD input dose ratio (1.5 times). The total Ka,r was 1.1 times higher than ICM-DSA. Therefore, the total Ka,r did not depend on the FPD input dose ratio. Multiple regression analyses of ICM-DSA and CO2-DSA showed significantly strong positive rank correlations and linear regressions for total Ka,r and total KAP (Figure 4;[ICM-DSA]:r = 0.966, R2 = 0.971;[CO2-DSA]:r = 0.806, R2 = 0.6966). There was also a positive rank correlation and linear regression between the total Ka,r and the total number of acquisitions ([ICM-DSA]:r = 0.5584, R2 = 0.3118;[CO2-DSA]:r = 0.672, R2 = 0.4511). The results of the analyses of each DSA image are presented below. The Ka,r for ICM-DSA was 31.1 ± 44.4 mGy, and that for CO2-DSA was 73.4 ± 88.7 mGy (p < 0.001). The KAP for ICM-DSA was 5.8059 ± 9.9655 Gy cm2 and that for CO2-DSA was 1.41926 ± 1.59652 Gy cm2 (p < 0.001). Multiple regression analysis showed that body weight (p < 0.05) and BMI (p < 0.05) significantly influenced Ka,r in the ICM-DSA group, but not in the CO2-DSA group. Field of views were 39.0 ± 72.4 cm for ICM-DSA and 38.1 ± 68.6 cm for CO2-DSA, with no significant difference between the two groups (p = 0.36). Frame rates were 6.1 ± 0.6 frame/s for ICM-DSA and 6.1 ± 3.6 frame/s for CO2-DSA, with no significant difference between the two groups (p = 0.86).
Welch’s t-test demonstrated a correlation between Ka,r and fluoroscopy time for ICM-DSA (p < 0.001), but not for CO2-DSA (p = 0.87). No significant differences were found between the ICM-DSA and CO2-DSA groups in terms of other factors.
Multiple regression analysis of X-ray factorsThe Ka,r and tube potential showed a significant rank correlation and linear regression for both ICM-DSA and CO2-DSA. Compared with Ka,r, there were significant differences in tube potentials (Figure 5;[ICM-DSA]:r = 0.592, R2 = 0.829, p < 0.001;[CO2-DSA]:r = 0.686, R2 = 0.4701;p < 0.001), KAP (Figure 6;[ICM-DSA]:r = 0.777, R2 = 0.6038, p < 0.001;[CO2-DSA]:r = 0.949, R2 = 0.8998;p < 0.001), and spectral shaping filters (p < 0.05) in both groups. There were no significant differences in patient-to-FPD distance, tube current, or frame rate between the groups. The tube potential, spectral shaping filter, tube current, and pulse width varied with the FPD input dose rate (Table 4).
The linear correlations between Ka,r and the number of ICM-DSA and CO2-DSA images varied among the three following types of doses (Figure 7):Ka,r > 1.2 times the number of DSA images (type A:higher dose required per DSA image);Ka,r < 0.8 times the number of DSA images (type C:lower dose required per DSA image);and the intermediate type (type B). Linear correlations were observed for types A and B, but not for type C. Correlation coefficients were observed for type A ([ICM-DSA] r = 0.670, R2 = 0.4491, a = 0.2172, p < 0.001;[CO2-DSA] r = 0.792, R2 = 0.628, a = 0.1633, p < 0.001), type B ([ICM-DSA] r = 0.989, R2 = 0.9348, a = 1.2084, p = 0.570;[CO2-DSA] r = 0.967, R2 = 0.9775, a = 0.9875, p = 0.887), and type C ([ICM-DSA] r = 0.087, R2 = 0.0075, a = 0.3556, p < 0.001;[CO2-DSA] r = 0.288, R2 = 0.0827, a = 0.6767, p < 0.001). The ratios of ICM-DSA to CO2-DSA images were 39.8:60.2 for type A, 71.0:29.0 for type B, and 72.0:28.0 for type C. Significant differences were observed between ICM-DSA and CO2-DSA for type A and C lesions. The number of DSA images was analyzed using multiple regression analysis. There were significant differences in the pulse width between ICM-DSA and CO2-DSA (p < 0.001).
Results of spatial scattered radiation dose rateThe spatially scattered radiation dose rates of the physicians in the standing position are listed in Table 5. In the protocol set up by the vendor, the rate for ICM-DSA and CO2-DSA (V) was 6.77 ± 0.12 mSv/h and 4.04 mSv/h, respectively. Furthermore, the FPD input dose rates in the CO2-DSA were 7.70 ± 0.22 mSv/h (I), 1.92 ± 0.04 mSv/h (III) and 2.70 ± 0.04 mSv/h (IV). The spatially scattered dose rates varied linearly. The FPD incident dose rate was the same and was 11.8 ± 0.34 mSv/h (II), where only the tube potential varied.
ICM-DSA and CO2-DSA yielded 49 and 21 patients, respectively. The number of cases of CO2-DSA was lower than that of ICM-DSA;however, this was considered reasonable for evaluation because only one physician performed the IR and excluded cases performed with other angiographic equipment. The interpretation of these results involves a comparison of the doses used in each protocol employed at Fukushima Medical University Hospital. Although the FPD input dose rates for the ICM-DSA and CO2-DSA protocols differed, no differences in the total Ka,r and total KAP were observed. Previous IR studies have also reported strong correlations between total Ka,r and total KAP. In this study, the data obtained were considered unbiased33).
In some cases, coronary angiography was conducted in conjunction with the diagnosis and IR for LEAD, or both ICM and CO2 were used;therefore, the total Ka,r and total KAP were not equivalent to the procedure dose for LEAD. As LEAD can be associated with coronary artery disease, this study included patients in whom coronary angiography was performed during LEAD treatment. There were significant differences in the tube potential and spectral shaping filter between the ICM-DSA and CO2-DSA groups because the angiographic system was equipped with an ADRC system, and the LEAD thickness varied greatly depending on the affected area. The body thickness was greater in the abdominal aorta and common iliac artery regions than in the popliteal artery region. ICM-DSA was used in the abdominal aorta and common iliac artery region in 11.4% of the cases and in the popliteal artery in 37.5% of the cases, whereas CO2-DSA was used in the abdominal aorta and common iliac artery in 27.0% of the cases and in the popliteal artery in 14.9% of the cases. The input dose rate of CO2-DSA was set 1.5 times higher than that of ICM-DSA. The spatial scattered radiation dose rate was 1.68 times higher;thus, the radiation exposure of the physician depends to some extent on the FPD input dose rate. Furthermore, significant differences in tube potential and spectral shaping filter values between ICM-DSA and CO2-DSA may be due to differences in the FPD input dose rate and tube potential settings, as well as differences in body thickness. The current dataset included cases in which there were not enough spectral shaping filters, and the use of appropriate spectral shaping filters could result in much lower doses. The spectral shaping filter should be determined by the desired vessel diameter, body thickness, and X-ray tube volume to control various X-ray quality-determining factors; however, at least 0.1 mm Cu should be inserted to reduce the incident skin dose. It is possible to set appropriate spectral shaping filters by measuring patient and physician radiation exposure when changing protocols. Multiple regression analysis showed that body weight and BMI significantly influenced Ka,r in the ICM-DSA group, but not in the CO2-DSA group. Since CO2-DSA is set to a higher tube voltage (81 kV), it is thought that it is less susceptible to the effects of changes in BMI. There were cases where a spectral shaping filter was not used due to the thickness of the imaging acquisition area, and in these cases, there was significant variation in tube potential. This was because the spectral shaping filter was automatically switched to 0 mmCu to maintain the FPD incident dose rate. For protocol optimization, a minimum of 0.1 mmCu spectral shaping filter is recommended for all regions. When used in peripheral arteries, consideration of a thicker spectral shaping filter may result in lower Ka,r and KAP.
Normalized KAP was higher for CO2-DSA than for ICM-DSA;however, KAP for each DSA series was significantly lower for CO2-DSA than for ICM-DSA. Since normalized KAP includes fluoroscopy, the effects of exposure from fluoroscopy cannot be overlooked in CO2-DSA. KAP for each DSA series does not include fluoroscopy, but it is greatly affected by the number of frames within a series. Since ICM-DSA had significantly more frames than CO2-DSA, it is thought that KAP was significantly lower for CO2-DSA than for ICM-DSA.
There were three doses for which the number of DSA images correlated with Ka,r regardless of the contrast medium. More than 40% of type A cases involved the abdominal aorta or common iliac artery, whereas more than 95% of type C cases involved peripheral arteries distal to the external iliac artery. Therefore, types A, B, and C were affected by differences in contrast vessels and body thickness. When used in the type A area, it is thought that the dose per image increases due to the use of a thin spectral shaping filter in order to maintain the FPD input dose. The use of a wide pulse width may have had a strong influence on Ka,r in type A, whereas the other types used narrow pulse widths, which did not affect Ka,r. The tube current did not affect Ka,r, suggesting that the pulse width controlled the X-ray dose. In addition, CO2-DSA used significantly wider pulse widths. ICM-DSA was mixed with blood, whereas CO2-DSA replaced blood with CO2 gas for the contrast effect. Therefore, a wide pulse width was used in CO2-DSA to depict more of the CO2 bubble dynamics in a single DSA image. When the DSA dose was optimized such that type A became type C, there was no significant difference between the contrast media, making it necessary to focus on image quality assurance. Consequently, because ICM-DSA and CO2-DSA protocols differ, dose optimization must be performed for each contrast medium.
Varying the CO2-DSA conditions at a height of 150 cm resulted in spatially scattered radiation doses (Table 5). The spatially scattered radiation dose at the position of the physician was very high. Physicians are often unaware of the set X-ray conditions and FPD incident dose rates for the protocols used in the procedure. Therefore, the dose of the physician may increase depending on the DSA protocol and imaging site. Physicians should maintain an appropriate distance from the X-ray tube, wear goggles and gloves, effectively use ceiling-suspended protective plates, select appropriate pulse and frame rates, apply an irradiation field aperture, avoid unnecessary imaging, and substitute fluoroscopic imaging functions where possible to minimize direct exposure to X-rays34-38). In particular, ceiling-suspended protective plates and goggles are effective at reducing the eye lens dose34,36,39-41).
The limitations of this study include the fact that it is a study of data from only one DSA system. Besides, this study is a comparison between specific protocols at a single facility, and caution is required when generalizing these results to other protocols or devices. In addition, this study did not consider the experiences of physicians or radiological technologists. Moreover, the FPD incident dose rate was assessed only for five patterns, and more detailed studies are required in the future. Similarly, angiography for LEAD is performed over a very wide range, from the trunk to the periphery of the lower limbs, and includes areas of varying thickness. In this study, the spatial scattering dose rate was evaluated using only 20 cm thick PMMA, and therefore, the evaluation of other body thicknesses was not performed, which is a limitation of this study. Furthermore, only DSA results for the LEAD were examined. Additionally, one physician decided on the selection of vessels to be imaged and the timing of imaging. Finally, image quality was not evaluated, and the equipment was one generation old.
Numbers of each artery acquired at DSA
DSA, digital subtraction angiography;ICM-DSA, iodine contrast medium digital subtraction angiography; CO2-DSA, carbon dioxide digital subtraction angiography
Results of multiple regression analysis
ICM-DSA, iodine contrast medium-digital subtraction angiography;CO2-DSA, carbon dioxide digital subtraction angiography;Ka,r, air kerma at the patient entrance reference point;KAP, the total kerma area product;BMI, body mass index
Comparison of total air kerma at patient entrance reference point (Ka,r) per acquisition between iodine contrast medium-digital subtraction angiography (ICM-DSA) and carbon dioxide DSA (CO2-DSA) n.s., not significant
Comparison of total kerma area product (KAP) per acquisition between iodine contrast medium-digital subtraction angiography (ICM-DSA) and carbon dioxide DSA (CO2-DSA)
Linear correlations between total air kerma at patient entrance reference point (Ka,r) and kerma area product (KAP) for iodine contrast medium-digital subtraction angiography (ICM-DSA) and carbon dioxide DSA (CO2-DSA)
Linear correlations between air kerma at patient entrance reference point (Ka,r) and tube potential for iodine contrast media-digital subtraction angiography (ICM-DSA) and carbon dioxide DSA (CO2-DSA)
Linear correlations between air kerma at patient entrance reference point (Ka,r) and kerma area product (KAP) for iodine contrast medium-digital subtraction angiography (ICM-DSA) and carbon dioxide DSA (CO2-DSA)
Results of comparison of X-ray factors
ICM-DSA, iodine contrast medium-digital subtraction angiography;CO2-DSA, carbon dioxide digital subtraction angiography;Ka,r, air kerma at the patient entrance reference point;KAP, the total kerma area product;n.s., no significant
Linear correlations between air kerma at patient entrance reference point Ka,r and number of digital subtraction angiography images for iodine contrast medium-digital subtraction angiography (ICM-DSA) and carbon dioxide digital subtraction angiography (CO2-DSA)
Results of spatial scattered radiation dose rate
ICM-DSA, iodine contrast medium-digital subtraction angiography; CO2-DSA, carbon dioxide digital subtraction angiography
The patient dose was retrospectively evaluated using the protocol used in LEAD. Physicians must be aware of radiation protection when conducting hand injections with CO2. An attempt should be made to optimize the exposure and image quality by creating DSA protocols for the most frequently used areas. Considering the need for high dose settings in CO2-DSA, we recommend radiation protection measures for surgeons (e.g., increased use of protective equipment).
We thank Susan Furness, Ph.D., from Edanz (https://jp.edanz.com/ac) for editing the manuscript.
All authors have no conflicts of interest to declare.
The study protocol was approved by the Institutional Review Board of Fukushima Medical University (REC2024-116).
The requirement for written informed consent was waived as direct patient involvement was not required.
The data from this study are available from the corresponding author upon reasonable request.
No funding was obtained for this study.
Kakuta and Hasegawa designed and conceived this study. K Kakuta, M Naruse, S Nemoto, and M Ikeda collected data. K Kakuta and K Chida analyzed and interpreted the results, drafted the manuscript, and performed the statistical analyses. All the authors have read and approved the final version of the manuscript.
