- Email: [email protected]
Monoenergetic Imaging of Dual-energy CT Reduces Artifacts from Implanted Metal Orthopedic Devices in Patients with Factures
Monoenergetic Imaging of Dual-energy CT Reduces Artifacts from Implanted Metal Orthopedic Devices in Patients with Factures Changsheng Zhou, BS, Yan E. Zhao, MS, Song Luo, BS, Hongyuan Shi, MS, Lin li, BS, Ling Zheng, BS, Long Jiang Zhang, MD, PhD, Guangming Lu, MD Rationale and Objectives: The purpose of this study was to optimize photon energy setting to reduce metal artifact of computed tomography (CT) images from implanted metal orthopedic devices in patients with fractures with monoenergetic imaging of dual-energy CT. Materials and Methods: This study included 47 patients with factures who underwent metal orthopedic device implanting. After dualenergy CT scan, monoenergetic software was used to postprocess with the following six photon energies: 40 kiloelectron-voltage (keV), 70 keV, 100 keV, 130 keV, 160 keV, and 190 keV. Two radiologists evaluated and rated the reformatted images with six different photon energies and average weighted 120 kVp images according to 4-score scale. The Wilcoxon rank-sum test was used to compare image quality scores for total, internal, and external metal orthopedic devices. Interreader agreement for image quality scoring was calculated. Results: Monoenergetic imaging of dual-energy CT improved the quality of CT images in the fracture patients with metal orthopedic devices compared to the average weighted 120 kVp images for the total, external, and internal metal orthopedic devices (all P values < .01). Optimal keV setting with the lowest metal artifact was 130 keV for total, internal, and external metal orthopedic devices. Good interreader agreement was found for the evaluation of image quality for total, internal, and external metal orthopedic devices. Conclusions: Monoenergetic imaging of dual-energy CT improves quality of CT images in patients with metal orthopedic devices after fracture. Reformatted images at 130 keV have the optimal quality for total, internal, and external metal orthopedic devices. Key Words: Tomography; X-ray computed; fracture; monoenergetic imaging; dual energy. ªAUR, 2011
M
ultidetector computed tomography (CT) with short scan times and isotropic spatial resolution has been widely used in the preoperative and postoperative evaluation of bony lesions; however, the quality of CT images can still be markedly deteriorated by the presence of metal objects in the field of view. Metal artifacts, mostly from quantum noise, scattered radiation, and beam hardening (1), influence image quality by reducing contrast and by obscuring details, thus impairing the detectability of structures of interest; in the worst case, this can make a diagnosis impossible (1–8). It is, therefore, of great importance to reduce these deteriorating artifacts to a minimum for better diagnosis.
Acad Radiol 2011; 18:1252–1257 Department of Medical Imaging, Jinling Hospital, Clinical School of Medical College, Nanjing University, 305 Zhongshan East Road, Xuanwu District, Nanjing, Jiangsu Province, 210002, China (C.Z., Y.E.Z., S.L., H.S., L.l., L.Z., L.J.Z., G.L.). Received January 16, 2011; accepted May 18, 2011. Address correspondence to: L.J.Z. e-mail: [email protected]; or G.L. e-mail: [email protected] ªAUR, 2011 doi:10.1016/j.acra.2011.05.009
1252
Many authors have recommended using higher milliampere second and peak kilovoltage techniques, the smallest collimator size, soft-tissue reconstruction filters, or extended CT Hounsfield unit scale to decrease CT artifacts arising from metal implants; however, these methods can deliver the patients higher radiation dose or decreased spatial resolution or even unavailable (9,10). Although many methods for metal artifact reduction in the raw data have been proposed, most algorithms did not provide sufficient quality to be accepted for clinical application (1,7). Methods for metal artifact reduction based on postreconstruction data have rarely been investigated (1–8). A dual-energy CT (DECT) technique, one image postprocessing method after image reconstruction (11), was used to reduce the beam-hardening artifact in CT in the 1980s (12). However, the dual-energy CT technique has not been used widely for clinical indications because of lower spatial resolution, unstable CT numbers, and insufficient tube currents at low tube voltages of the early CT scanners (13). Recently developed dual-source CT scanners with two orthogonally mounted detectors and tubes arrays operating simultaneously at different tube potentials (80 kVp and 140 kVp) allow for DECT
Academic Radiology, Vol 18, No 10, October 2011
acquisitions with minimal patient motion registration artifacts and were used to improve the detection of pulmonary embolism (14,15) and cardiovascular diseases (16,17), and to characterize the composition of kidney stones (18). Herein, we report our preliminary results of the simulated monoenergetic imaging of DECT in reducing the effect of metal artifact on the adjacent bone and soft tissue in 47 patients with implanted metal orthopedic devices after factures. MATERIALS AND METHODS Patient Population
This study had local institutional review board approval. Fortyseven consecutive patients (27 males and 20 females, age range 19–83 years, mean age 45 14 years) were included into this study between May and November 2010. All patients gave informed consent. Inclusion criteria are that any patients had fractures in the variable locations, undergoing internal (n = 36) or external implanted metal devices (n = 9), or internal and external implanted metal devices (n = 2); patients younger than 18 years old and pregnant or morbidly obese patients (body mass index >40 kg/m2 ) were excluded. Of 47 patients, 40 patients had 40 implanted metal devices, whereas 7 patients had 14 implanted metal devices. Of seven patients with multiple implanted metal devices, two patients had both external and internal implanted metal devices (external implanted metal devices in radius and internal implanted metal device in humerus, n = 1; internal implanted metal device in femur and external implanted metal device in tibiofibula, n = 1) and three patients had two internal implanted metal devices (tibia and fibula, n = 2; humerus and radius, n = 1); the remaining two patients had two external implanted metal devices (radius and humerus, n = 1; femur and tibiofibula, n = 1). Locations, numbers, and types of implanted metal devices are shown in Table 1. Dual-energy CT Imaging
DECT was performed using a dual-source CT system (SOMATOM Definition, Siemens Medical Solutions, Forchheim, Germany). The patients were centrally placed in the scanner to ensure that the entire thorax was covered with the smaller field of view of the second tube detector array; the size of the field of view of the second tube detector array was 260 mm. DECT scan was performed after scout with the following parameters: tube voltages of 80 kVp and 140 kVp and reference tube current of 468 mA and 86 mA for the two x-ray tubes, rotation time of 0.5 seconds, detector collimation of 14 1.2 mm, pitch of 0.8, and field of view of 260 mm. The range of dose length product was 36–160 mGy/cm for DECT scan. Image Postprocessing and Analysis
From the raw spiral projection data of both tubes, images were automatically reconstructed to three groups of datasets:
MONOENERGETIC IMAGING OF DUAL-ENERGY CT
TABLE 1. Locations, Numbers, and Types of Metal Implanted Metal Devices in this Study
Locations Lower extremity Upper extremity Cervical vertebra Lumbar vertebra
Internal Implanted External Implanted metal Device Metal Device Total 18 7 9 7
10 3 0 0
28 10 9 7
Data presented in this table are numbers of implanted metal devices.
80 kVp, 140 kVp, and average weighted images with decomposition of 0.3 (30% density information from 80 kVp scan and 70% from the 140 kVp scan) with slice thickness of 0.75 mm and interval of 0.50 mm (14–17). Then all data were transferred to a commercial workstation (Syngo VE32E, Siemens Medical Solutions). Average weighted 120 kV images from both tubes (80 kVp and 140 kVp) were transferred to a commercially available workstation (Syngommvvp VE32E, Siemens AG Medical Solutions, Forchheim, Germany) to obtain multiplanar reformation in the axial, coronal, and sagittal planes with slice thickness of 5 mm and volume rendering in different projections for diagnostic reading and image analysis. Slice thickness of 5 mm is a routine protocol used in our hospital to reduce the number of printed films. Both 80 kVp and 140 kVp data sets were transferred to another workstation (Syngommwp VE36A, Siemens AG, Medical Solutions) equipped with the monoenergetic software that can extract the monoenergetic images at the arbitrary photon energies ranging from 40 keV to 190 keV. Six datasets at 40 keV, 70 keV, 100 keV, 130 keV, 160 keV, and 190 keV were postprocessed and saved, then multiplanar reformation in the axial, coronal, and sagittal planes with slice thickness of 5 mm in different projections were performed for image analysis of the monoenergetic CT images. All images were evaluated independently by two musculoskeletal radiologists (C.Z. and S.L., with 8 and 3 years of experience in radiology reading, respectively) with a four-point scale: 1 (poor) was defined as prominent artifacts influencing the evaluation of the adjacent bony structures and soft tissues. A score of 2 (moderate) was defined as moderate artifacts slightly influencing the evaluation of the adjacent bony structures and soft tissues. Score 3 (good) was defined as mild artifacts, slightly not influencing the evaluation of the adjacent bony structures and soft tissues. Score 4 (excellent) was defined as no artifacts and clear visualization of the adjacent bone trabecula and soft tissues. Only poor image quality was regarded as nonassessable. The readers were instructed to evaluate the images quality based on trabecular bone definition 1 cm from the metal implant and degree of streak artifact. Spinal canal abnormalities, perihardware fluid collections, hardware fracture, and presence of loosening were also recorded if present. Both readers were blinded to the monoenergetic value of the image. Discrepancies between the 1253
ZHOU ET AL
Academic Radiology, Vol 18, No 10, October 2011
TABLE 2. Comparison of Image Quality of Metal Fixation Among Different Photon Energies (n) Score
40 keV
70 keV
Total 4 0/0 0/0 3 0/0 2/6 2 0/0 51/47 1 54/54 1/1 Rank 28.0/28.0 125.7/119.4 External implanted metal device 4 0/0 0/0 3 0/0 0/1 2 0/0 13/12 1 13/13 0/0 Rank 7.0/7.0 33.5/29.9 Internal implanted metal device 4 0/0 0/0 3 0/0 2/5 2 0/0 38/35 1 41/41 1/1 Rank 21.5/21.5 92.7/89.9
c2 value
P Value
0/0 10/13 44/41 0/0 148.9/139.2
0.671
<.01
2/0 8/11 3/2 0/0 64.5/58.7
0/0 0/0 13/13 0/0 33.5/27.0
0.700
<.01
0/0 33/38 8/3 0/0 176.7/175.4
0/0 10/13 31/28 0/0 115.6/111.6
0.661
<.01
100 keV
130 keV
160 keV
190 keV
120 kVp*
16/20 29/30 9/4 0/0 271.4/278.0
14/14 34/39 6/1 0/0 276.0/273.4
0/5 42/48 12/1 0/0 235.2/254.7
2/0 41/49 11/5 0/0 241.4/233.9
0/0 8/13 5/0 0/0 56.0/64.5
3/2 7/11 3/0 0/0 66.0/68.4
0/1 10/12 3/0 0/0 61.6/66.5
16/20 21/17 4/4 0/0 217.0/215.0
11/12 27/28 3/1 0/0 210.4/205.8
0/4 32/36 9/1 0/0 174.1/188.9
*Average weighted 120 kV data sets; data presented in this table indicate results for reader 1 and reader 2.
two readers were resolved by consulting in consensus. Identical window-level settings (window width, 1500; window level, 500 for bone window; soft-window width, 350; window level, 45 for soft-tissue window) was used to evaluate image quality in this study. Statistical Analysis
Statistical analysis was performed using the software SPSS version 13.0 (SPSS Inc, Chicago, IL). Quantitative variables were expressed as mean (SD) and categorical variables as frequencies or percentages. A Kruskal-Wallis test was used to compare the quality of CT images for the total, internal, and external metal orthopedic devices among different keV settings and the average weighted 120 kVp data. Kappa statistics was also calculated to quantify interreader variability for the evaluation of image quality of internal, external, and whole implanted metal orthopedic devices. Kappa values less than 0.20 were interpreted as poor agreement, 0.21–0.40 as fair agreement, 0.41–0.60 as moderate, 0.61–0.80 as good, and 0.81–1.00 as very good agreement. P values less than .05 were regarded as statistically significant.
130 keV for readers 1 and 2 versus 0% and 0% images with score $3 at the average weighted 120 kVp for readers 1 and 2; P < .01), and for the internal metal orthopedic devices (93% and 98% images with score $3 at 130 keV for readers 1 and 2 versus 0% and 0% images with score $3 at the average weighted 120 kVp for readers 1 and 2; P < .01). Table 2 illustrated the numbers of image quality score, rank, and statistical values from two readers for the simulated monoenergetic imaging at the six different keV settings and at the average weighted 120 kVp. Both readers had good agreement (kappa values = 0.700, 0.661, and 0.671 for the evaluation of CT images with the external, internal, and whole implanted metal orthopedic devices; all P values less than .01) for the evaluation of the quality of CT images. No spinal canal abnormalities, perihardware fluid collections, hardware fracture, or presence of loosening were observed. Figures 1 and 2 illustrated representative images in which beam-hardening artifact was reduced when the simulated monoenergetic imaging at 100–130 keV extracted from DECT data was performed compared with the average weighted 120 kVp images.
DISCUSSION RESULTS Monoenergetic imaging of DECT improved the quality of CT images in patients with metal orthopedic devices after fracture compared to the average weighted 120 kVp images for the total metal orthopedic devices (89% and 98% images with score $3 at 130 keV for readers 1 and 2 versus 19% and 24% images with score $3 at the average weighted 120 kVp for readers 1 and 2; P < .01), for the external metal orthopedic devices (77% and 100% images with score $3 at 1254
Our study indicated the ability of the simulated monoenergetic images extracted from DECT data to reduce metal artifact of CT images in the patients with implanted metal orthopedic devices after fractures; 130 keV is the optimal photon energy setting with the lowest metal artifact for the total, internal, and external implanted metal orthopedic devices. Methods of CT metal artifact reduction have been investigated, with its advantages and disadvantages for each type of
Academic Radiology, Vol 18, No 10, October 2011
MONOENERGETIC IMAGING OF DUAL-ENERGY CT
Figure 1. Representative images in a patient with internal implanted metal orthopedic devices in cervical vertebra. Figures a-g are reformatted monoenergetic computed tomography images at 40 keV, 70 keV, 100 keV, 130 keV, 160 keV, 190 keV, and average weighted 120 kVp. Beam hardening artifact is reduced at 100 keV and 130 keV than that at the average weighted 120 kVp image.
metal implant and CT data acquisition configuration in terms of flexibility, computational efficiency, and image quality in the corrected image (5). Adaptive filtering methods can reduce the streaking artifacts caused by photon starvation, but these methods cannot correct for the severe data inconsistencies caused by metal in the presence of highly attenuating metal (5). The key idea of the iterative method lies in the reconstruction of the CT image using only those noncorrupted projections while discarding those projections affected by the metal objects. Compared with the traditional filtered
back-projection algorithm, iterative methods are more robust in dealing with incomplete projections caused by metallic implants. However, it can achieve only a suboptimal solution, its convergence is sensitive to the initialization conditions, and it is computationally expensive for clinical applications (8). The major advantage of the projection modification method is its simplicity. In the projection modification approach, the metal shadows in the raw projection data caused by the x-ray passing through the metallic implants are first segmented and then replaced using some estimated values. Though the 1255
ZHOU ET AL
Academic Radiology, Vol 18, No 10, October 2011
Figure 2. Representative images in a patient with hip endoprosthesis. (a) Average weighted 120 kVp and (b) 130 keV soft-tissue window images show the left hip endoprosthesis; artifact of (b) decreased compared with (a). (c) Average weighted 120 kVp and (d) 130 keV bone window images show better image quality of (d) than that of (c); the image quality was rated as moderate.
projection modification method does not require expensive computation, a major challenge lies in the accurate detection of metal shadows in the raw projection data. Although variable algorithms (2,3) were investigated, the methods were shown to be impractical for clinical applications or difficult to implement (8). In this study, the ability of the simulated monoenergetic imaging from DECT data to reduce metal artifact of CT images in the patients with implanted metal orthopedic devices after fractures was investigated. At DECT, x-ray attenuation coefficient of the objects measured from simultaneously acquired 80 kVp and 140 kVp images was transformed into density images of a pair of the basic objects with the same attenuation coefficient as the previously mentioned objects, then linear computation equation was performed to generate the monoenergetic images ranging from 40 keV to 190 keV. The monoenergetic CT images of the arbitrary keV at this keV range can be extracted. In this study, we extracted six series of the monoenergetic CT images at 6 keV settings with an interval of 30 keV. We found the simulated monoenergetic imaging from DECT data was able to reduce metal artifact of CT images in the patients with implanted metal orthopedic devices after fractures. A value of 130 keV was the optimal setting for CT metal artifact reduction from the total, internal, and external 1256
implanted metal orthopedic devices. Our findings were in contrast to the results of one previous study in which the authors found noise in the monoenergetic images was lowest around 70–80 keV in the body and skull examinations (11). We believe the interested organs studied between both studies can result in the difference. One recently published work (19) indicated 105 keV and optimal energy provided the superior diagnostic image quality. In Bamberg et al’s study, monoenergetic DECT can reduce metal artefacts and may significantly enhance diagnostic value in the evaluation of metallic implants. In this study, a good interreader agreement was also found for the evaluation of image quality of reformatted monoenergetic CT images at different keV settings. This indicated the potential of the simulated monoenergetic imaging from DECT data in the clinical setting. In contrast to most other algorithms for the metal artifact reduction in the raw data, the dual-energy processing is performed after image reconstruction rather than on projection data before image reconstruction (12). This advantage of the simulated monoenergetic imaging of the DECT technique makes the radiologists to reduce the metal artifact of CT images in the busy clinical practice by using the dualenergy software equipped in the dedicated workstation, although it cannot completely eliminate the artifact (19). Thus, the DECT monoenergetic imaging for metal artifact
Academic Radiology, Vol 18, No 10, October 2011
reduction can be widely used in the clinics with dual source CT scanners. In addition, the range of dose-length product was 36–160 mGy$cm for DECT scan in this study, which was not higher than the dose of singl-energy CT scan in our experience because direct comparison of radiation dose between single-energy and DECT scans was not available or ethically allowable. We acknowledge the following study limitations. First, the small size of the study limits broad generalization of our findings, which deserve further study in a large cohort. However, the initial study demonstrated the ability of simulated monoenergetic imaging from DECT data to reduce metal artifact of CT images in the patients with implanted metal orthopedic devices after fractures. Second, we performed image postprocessing with 30 keV interval rather than 10 keV or 5 keV interval for the evaluation of metal artifact reduction of reformatted monoenergetic CT images, which can cause suboptimal selection of keV setting for metal artifact reduction. Thus, a large-cohort, detailed study is needed to resolve this issue. Third, the monoenergetic imaging in this study was used to postprocess the dual-energy CT data; the technique may be limited to the patients who underwent dual-source, DECT acquisition. In conclusion, this study shows that the monoenergetic imaging of DECT improves image quality of CT in the patients with implanted metal orthopedic devices compared with the average weighted 120 kVp image. The optimal image quality with the lowest metal artifact can be extracted at 130 keV for the total, external, and internal implanted metal orthopedic devices compared with the average weighted 120 kVp images.
MONOENERGETIC IMAGING OF DUAL-ENERGY CT
4.
5.
6.
7.
8.
9.
10.
11.
12.
13. 14.
15.
16.
17.
REFERENCES 1. Watzke O, Kalender WA. A pragmatic approach to metal artifact reduction in CT: merging of metal artifact reduced images. Eur Radiol 2004; 14: 849–856. 2. Kalender WA, Hebel R, Ebersberger J. Reduction of CT artifacts caused by metallic implants. Radiology 1987; 164:576–577. 3. Yazdia M, Gingras L, Beaulieu L. An adaptive approach to metal artifact reduction in helical computed tomography for radiation therapy treatment
18.
19.
planning: experimental and clinical studies. Int J Radiat Oncol Biol Phys 2005; 62:1224–1231. Liu PT, Pavlicek WP, Peter MB, et al. Metal artifact reduction image reconstruction algorithm for CT of implanted metal orthopedic devices: a work in progress. Skeletal Radiol 2009; 38:797–802. Yu L, Li H, Mueller J, et al. Metal artifact reduction from reformatted projections for hip prostheses in multislice helical computed tomography techniques and initial clinical results. Invest Radiol 2009; 44: 691–696. Prell D, Kyriakou Y, Kachelrieß M, et al. Reducing metal artifacts in computed tomography caused by hip endoprostheses using a physicsbased approach. Invest Radiol 2010; 45:747–754. Prell D, Kyriakou Y, Struffert T, et al. Metal artifact reduction for clipping and coiling in interventional c-arm CT. AJNR Am J Neuroradiol 2010; 31: 634–639. Zhang Y, Zhang L, Zhu XR, et al. Reducing metal artifacts in con-beam CT images by preprocessing projection data. Int J Radiat Oncol Biol Phys 2007; 67:924–932. Lee MJ, Kim S, Lee SA, et al. Overcoming artifacts from metallic orthopedic implants at high-field-strength MR imaging and multi-detector CT. Radiographics 2007; 27:791–803. Link TM, Berning W, Scherf S, et al. CT of metal implants: reduction of artifacts using an extended CT scale technique. J Comput Assist Tomogr 2000; 24:165–172. Hemmingsson A, Jung B, Ytterbergh C. Dual energy computed tomography: simulated monoenergetic and material-selective imaging. J Comput Assist Tomogr 1986; 10:490–499. Fletcher JG, Takahashi N, Hartman R, et al. Dual-energy and dual-source CT: is there a role in the abdomen and pelvis. Radiol Clin N Am 2009; 47: 41–57. Johnson TR, Krauss B, Sedlmair M, et al. Material differentiation by dual energy CT: initial experience. Eur Radiol 2007; 17:1510–1517. Zhang LJ, Zhao YE, Wu SY, et al. Pulmonary embolism detection with dual-energy CT: experimental study of dual-source CT in rabbits. Radiology 2009; 252:61–70. Zhang LJ, Chai X, Wu SY, et al. Detection of pulmonary embolism by dual energy CT: correlation with perfusion scintigraphy and pathohistological findings in rabbits. Eur Radiol 2009; 19:2844–2854. Zhang LJ, Wu SY, Niu JB, et al. Dual energy CT angiography for the evaluation of intracranial aneurysms: image quality, radiation dose, and comparison with 3D rotational digital subtraction angiography. AJR Am J Roentgenol 2010; 194:23–30. Zhang LJ, Peng J, Wu SY, et al. Dual source dual energy CT of acute myocardial infraction: correlation with histopathological findings in a canine model. Invest Radiol 2010; 45:290–297. Thomas C, Heuschmid M, Schilling D, et al. Urinary calculi composed of uric acid, cystine, and mineral salts: differentiation with dual-energy CT at a radiation dose comparable to that of intravenous pyelography. Radiology 2010; 257:402–409. Bamberg F, Dierks A, Nikolaou K, et al. Metal artifact reduction by dual energy computed tomography using monoenergetic extrapolation. Eur Radiol 2011; 21:1424–1429.
1257
M
ultidetector computed tomography (CT) with short scan times and isotropic spatial resolution has been widely used in the preoperative and postoperative evaluation of bony lesions; however, the quality of CT images can still be markedly deteriorated by the presence of metal objects in the field of view. Metal artifacts, mostly from quantum noise, scattered radiation, and beam hardening (1), influence image quality by reducing contrast and by obscuring details, thus impairing the detectability of structures of interest; in the worst case, this can make a diagnosis impossible (1–8). It is, therefore, of great importance to reduce these deteriorating artifacts to a minimum for better diagnosis.
Acad Radiol 2011; 18:1252–1257 Department of Medical Imaging, Jinling Hospital, Clinical School of Medical College, Nanjing University, 305 Zhongshan East Road, Xuanwu District, Nanjing, Jiangsu Province, 210002, China (C.Z., Y.E.Z., S.L., H.S., L.l., L.Z., L.J.Z., G.L.). Received January 16, 2011; accepted May 18, 2011. Address correspondence to: L.J.Z. e-mail: [email protected]; or G.L. e-mail: [email protected] ªAUR, 2011 doi:10.1016/j.acra.2011.05.009
1252
Many authors have recommended using higher milliampere second and peak kilovoltage techniques, the smallest collimator size, soft-tissue reconstruction filters, or extended CT Hounsfield unit scale to decrease CT artifacts arising from metal implants; however, these methods can deliver the patients higher radiation dose or decreased spatial resolution or even unavailable (9,10). Although many methods for metal artifact reduction in the raw data have been proposed, most algorithms did not provide sufficient quality to be accepted for clinical application (1,7). Methods for metal artifact reduction based on postreconstruction data have rarely been investigated (1–8). A dual-energy CT (DECT) technique, one image postprocessing method after image reconstruction (11), was used to reduce the beam-hardening artifact in CT in the 1980s (12). However, the dual-energy CT technique has not been used widely for clinical indications because of lower spatial resolution, unstable CT numbers, and insufficient tube currents at low tube voltages of the early CT scanners (13). Recently developed dual-source CT scanners with two orthogonally mounted detectors and tubes arrays operating simultaneously at different tube potentials (80 kVp and 140 kVp) allow for DECT
Academic Radiology, Vol 18, No 10, October 2011
acquisitions with minimal patient motion registration artifacts and were used to improve the detection of pulmonary embolism (14,15) and cardiovascular diseases (16,17), and to characterize the composition of kidney stones (18). Herein, we report our preliminary results of the simulated monoenergetic imaging of DECT in reducing the effect of metal artifact on the adjacent bone and soft tissue in 47 patients with implanted metal orthopedic devices after factures. MATERIALS AND METHODS Patient Population
This study had local institutional review board approval. Fortyseven consecutive patients (27 males and 20 females, age range 19–83 years, mean age 45 14 years) were included into this study between May and November 2010. All patients gave informed consent. Inclusion criteria are that any patients had fractures in the variable locations, undergoing internal (n = 36) or external implanted metal devices (n = 9), or internal and external implanted metal devices (n = 2); patients younger than 18 years old and pregnant or morbidly obese patients (body mass index >40 kg/m2 ) were excluded. Of 47 patients, 40 patients had 40 implanted metal devices, whereas 7 patients had 14 implanted metal devices. Of seven patients with multiple implanted metal devices, two patients had both external and internal implanted metal devices (external implanted metal devices in radius and internal implanted metal device in humerus, n = 1; internal implanted metal device in femur and external implanted metal device in tibiofibula, n = 1) and three patients had two internal implanted metal devices (tibia and fibula, n = 2; humerus and radius, n = 1); the remaining two patients had two external implanted metal devices (radius and humerus, n = 1; femur and tibiofibula, n = 1). Locations, numbers, and types of implanted metal devices are shown in Table 1. Dual-energy CT Imaging
DECT was performed using a dual-source CT system (SOMATOM Definition, Siemens Medical Solutions, Forchheim, Germany). The patients were centrally placed in the scanner to ensure that the entire thorax was covered with the smaller field of view of the second tube detector array; the size of the field of view of the second tube detector array was 260 mm. DECT scan was performed after scout with the following parameters: tube voltages of 80 kVp and 140 kVp and reference tube current of 468 mA and 86 mA for the two x-ray tubes, rotation time of 0.5 seconds, detector collimation of 14 1.2 mm, pitch of 0.8, and field of view of 260 mm. The range of dose length product was 36–160 mGy/cm for DECT scan. Image Postprocessing and Analysis
From the raw spiral projection data of both tubes, images were automatically reconstructed to three groups of datasets:
MONOENERGETIC IMAGING OF DUAL-ENERGY CT
TABLE 1. Locations, Numbers, and Types of Metal Implanted Metal Devices in this Study
Locations Lower extremity Upper extremity Cervical vertebra Lumbar vertebra
Internal Implanted External Implanted metal Device Metal Device Total 18 7 9 7
10 3 0 0
28 10 9 7
Data presented in this table are numbers of implanted metal devices.
80 kVp, 140 kVp, and average weighted images with decomposition of 0.3 (30% density information from 80 kVp scan and 70% from the 140 kVp scan) with slice thickness of 0.75 mm and interval of 0.50 mm (14–17). Then all data were transferred to a commercial workstation (Syngo VE32E, Siemens Medical Solutions). Average weighted 120 kV images from both tubes (80 kVp and 140 kVp) were transferred to a commercially available workstation (Syngommvvp VE32E, Siemens AG Medical Solutions, Forchheim, Germany) to obtain multiplanar reformation in the axial, coronal, and sagittal planes with slice thickness of 5 mm and volume rendering in different projections for diagnostic reading and image analysis. Slice thickness of 5 mm is a routine protocol used in our hospital to reduce the number of printed films. Both 80 kVp and 140 kVp data sets were transferred to another workstation (Syngommwp VE36A, Siemens AG, Medical Solutions) equipped with the monoenergetic software that can extract the monoenergetic images at the arbitrary photon energies ranging from 40 keV to 190 keV. Six datasets at 40 keV, 70 keV, 100 keV, 130 keV, 160 keV, and 190 keV were postprocessed and saved, then multiplanar reformation in the axial, coronal, and sagittal planes with slice thickness of 5 mm in different projections were performed for image analysis of the monoenergetic CT images. All images were evaluated independently by two musculoskeletal radiologists (C.Z. and S.L., with 8 and 3 years of experience in radiology reading, respectively) with a four-point scale: 1 (poor) was defined as prominent artifacts influencing the evaluation of the adjacent bony structures and soft tissues. A score of 2 (moderate) was defined as moderate artifacts slightly influencing the evaluation of the adjacent bony structures and soft tissues. Score 3 (good) was defined as mild artifacts, slightly not influencing the evaluation of the adjacent bony structures and soft tissues. Score 4 (excellent) was defined as no artifacts and clear visualization of the adjacent bone trabecula and soft tissues. Only poor image quality was regarded as nonassessable. The readers were instructed to evaluate the images quality based on trabecular bone definition 1 cm from the metal implant and degree of streak artifact. Spinal canal abnormalities, perihardware fluid collections, hardware fracture, and presence of loosening were also recorded if present. Both readers were blinded to the monoenergetic value of the image. Discrepancies between the 1253
ZHOU ET AL
Academic Radiology, Vol 18, No 10, October 2011
TABLE 2. Comparison of Image Quality of Metal Fixation Among Different Photon Energies (n) Score
40 keV
70 keV
Total 4 0/0 0/0 3 0/0 2/6 2 0/0 51/47 1 54/54 1/1 Rank 28.0/28.0 125.7/119.4 External implanted metal device 4 0/0 0/0 3 0/0 0/1 2 0/0 13/12 1 13/13 0/0 Rank 7.0/7.0 33.5/29.9 Internal implanted metal device 4 0/0 0/0 3 0/0 2/5 2 0/0 38/35 1 41/41 1/1 Rank 21.5/21.5 92.7/89.9
c2 value
P Value
0/0 10/13 44/41 0/0 148.9/139.2
0.671
<.01
2/0 8/11 3/2 0/0 64.5/58.7
0/0 0/0 13/13 0/0 33.5/27.0
0.700
<.01
0/0 33/38 8/3 0/0 176.7/175.4
0/0 10/13 31/28 0/0 115.6/111.6
0.661
<.01
100 keV
130 keV
160 keV
190 keV
120 kVp*
16/20 29/30 9/4 0/0 271.4/278.0
14/14 34/39 6/1 0/0 276.0/273.4
0/5 42/48 12/1 0/0 235.2/254.7
2/0 41/49 11/5 0/0 241.4/233.9
0/0 8/13 5/0 0/0 56.0/64.5
3/2 7/11 3/0 0/0 66.0/68.4
0/1 10/12 3/0 0/0 61.6/66.5
16/20 21/17 4/4 0/0 217.0/215.0
11/12 27/28 3/1 0/0 210.4/205.8
0/4 32/36 9/1 0/0 174.1/188.9
*Average weighted 120 kV data sets; data presented in this table indicate results for reader 1 and reader 2.
two readers were resolved by consulting in consensus. Identical window-level settings (window width, 1500; window level, 500 for bone window; soft-window width, 350; window level, 45 for soft-tissue window) was used to evaluate image quality in this study. Statistical Analysis
Statistical analysis was performed using the software SPSS version 13.0 (SPSS Inc, Chicago, IL). Quantitative variables were expressed as mean (SD) and categorical variables as frequencies or percentages. A Kruskal-Wallis test was used to compare the quality of CT images for the total, internal, and external metal orthopedic devices among different keV settings and the average weighted 120 kVp data. Kappa statistics was also calculated to quantify interreader variability for the evaluation of image quality of internal, external, and whole implanted metal orthopedic devices. Kappa values less than 0.20 were interpreted as poor agreement, 0.21–0.40 as fair agreement, 0.41–0.60 as moderate, 0.61–0.80 as good, and 0.81–1.00 as very good agreement. P values less than .05 were regarded as statistically significant.
130 keV for readers 1 and 2 versus 0% and 0% images with score $3 at the average weighted 120 kVp for readers 1 and 2; P < .01), and for the internal metal orthopedic devices (93% and 98% images with score $3 at 130 keV for readers 1 and 2 versus 0% and 0% images with score $3 at the average weighted 120 kVp for readers 1 and 2; P < .01). Table 2 illustrated the numbers of image quality score, rank, and statistical values from two readers for the simulated monoenergetic imaging at the six different keV settings and at the average weighted 120 kVp. Both readers had good agreement (kappa values = 0.700, 0.661, and 0.671 for the evaluation of CT images with the external, internal, and whole implanted metal orthopedic devices; all P values less than .01) for the evaluation of the quality of CT images. No spinal canal abnormalities, perihardware fluid collections, hardware fracture, or presence of loosening were observed. Figures 1 and 2 illustrated representative images in which beam-hardening artifact was reduced when the simulated monoenergetic imaging at 100–130 keV extracted from DECT data was performed compared with the average weighted 120 kVp images.
DISCUSSION RESULTS Monoenergetic imaging of DECT improved the quality of CT images in patients with metal orthopedic devices after fracture compared to the average weighted 120 kVp images for the total metal orthopedic devices (89% and 98% images with score $3 at 130 keV for readers 1 and 2 versus 19% and 24% images with score $3 at the average weighted 120 kVp for readers 1 and 2; P < .01), for the external metal orthopedic devices (77% and 100% images with score $3 at 1254
Our study indicated the ability of the simulated monoenergetic images extracted from DECT data to reduce metal artifact of CT images in the patients with implanted metal orthopedic devices after fractures; 130 keV is the optimal photon energy setting with the lowest metal artifact for the total, internal, and external implanted metal orthopedic devices. Methods of CT metal artifact reduction have been investigated, with its advantages and disadvantages for each type of
Academic Radiology, Vol 18, No 10, October 2011
MONOENERGETIC IMAGING OF DUAL-ENERGY CT
Figure 1. Representative images in a patient with internal implanted metal orthopedic devices in cervical vertebra. Figures a-g are reformatted monoenergetic computed tomography images at 40 keV, 70 keV, 100 keV, 130 keV, 160 keV, 190 keV, and average weighted 120 kVp. Beam hardening artifact is reduced at 100 keV and 130 keV than that at the average weighted 120 kVp image.
metal implant and CT data acquisition configuration in terms of flexibility, computational efficiency, and image quality in the corrected image (5). Adaptive filtering methods can reduce the streaking artifacts caused by photon starvation, but these methods cannot correct for the severe data inconsistencies caused by metal in the presence of highly attenuating metal (5). The key idea of the iterative method lies in the reconstruction of the CT image using only those noncorrupted projections while discarding those projections affected by the metal objects. Compared with the traditional filtered
back-projection algorithm, iterative methods are more robust in dealing with incomplete projections caused by metallic implants. However, it can achieve only a suboptimal solution, its convergence is sensitive to the initialization conditions, and it is computationally expensive for clinical applications (8). The major advantage of the projection modification method is its simplicity. In the projection modification approach, the metal shadows in the raw projection data caused by the x-ray passing through the metallic implants are first segmented and then replaced using some estimated values. Though the 1255
ZHOU ET AL
Academic Radiology, Vol 18, No 10, October 2011
Figure 2. Representative images in a patient with hip endoprosthesis. (a) Average weighted 120 kVp and (b) 130 keV soft-tissue window images show the left hip endoprosthesis; artifact of (b) decreased compared with (a). (c) Average weighted 120 kVp and (d) 130 keV bone window images show better image quality of (d) than that of (c); the image quality was rated as moderate.
projection modification method does not require expensive computation, a major challenge lies in the accurate detection of metal shadows in the raw projection data. Although variable algorithms (2,3) were investigated, the methods were shown to be impractical for clinical applications or difficult to implement (8). In this study, the ability of the simulated monoenergetic imaging from DECT data to reduce metal artifact of CT images in the patients with implanted metal orthopedic devices after fractures was investigated. At DECT, x-ray attenuation coefficient of the objects measured from simultaneously acquired 80 kVp and 140 kVp images was transformed into density images of a pair of the basic objects with the same attenuation coefficient as the previously mentioned objects, then linear computation equation was performed to generate the monoenergetic images ranging from 40 keV to 190 keV. The monoenergetic CT images of the arbitrary keV at this keV range can be extracted. In this study, we extracted six series of the monoenergetic CT images at 6 keV settings with an interval of 30 keV. We found the simulated monoenergetic imaging from DECT data was able to reduce metal artifact of CT images in the patients with implanted metal orthopedic devices after fractures. A value of 130 keV was the optimal setting for CT metal artifact reduction from the total, internal, and external 1256
implanted metal orthopedic devices. Our findings were in contrast to the results of one previous study in which the authors found noise in the monoenergetic images was lowest around 70–80 keV in the body and skull examinations (11). We believe the interested organs studied between both studies can result in the difference. One recently published work (19) indicated 105 keV and optimal energy provided the superior diagnostic image quality. In Bamberg et al’s study, monoenergetic DECT can reduce metal artefacts and may significantly enhance diagnostic value in the evaluation of metallic implants. In this study, a good interreader agreement was also found for the evaluation of image quality of reformatted monoenergetic CT images at different keV settings. This indicated the potential of the simulated monoenergetic imaging from DECT data in the clinical setting. In contrast to most other algorithms for the metal artifact reduction in the raw data, the dual-energy processing is performed after image reconstruction rather than on projection data before image reconstruction (12). This advantage of the simulated monoenergetic imaging of the DECT technique makes the radiologists to reduce the metal artifact of CT images in the busy clinical practice by using the dualenergy software equipped in the dedicated workstation, although it cannot completely eliminate the artifact (19). Thus, the DECT monoenergetic imaging for metal artifact
Academic Radiology, Vol 18, No 10, October 2011
reduction can be widely used in the clinics with dual source CT scanners. In addition, the range of dose-length product was 36–160 mGy$cm for DECT scan in this study, which was not higher than the dose of singl-energy CT scan in our experience because direct comparison of radiation dose between single-energy and DECT scans was not available or ethically allowable. We acknowledge the following study limitations. First, the small size of the study limits broad generalization of our findings, which deserve further study in a large cohort. However, the initial study demonstrated the ability of simulated monoenergetic imaging from DECT data to reduce metal artifact of CT images in the patients with implanted metal orthopedic devices after fractures. Second, we performed image postprocessing with 30 keV interval rather than 10 keV or 5 keV interval for the evaluation of metal artifact reduction of reformatted monoenergetic CT images, which can cause suboptimal selection of keV setting for metal artifact reduction. Thus, a large-cohort, detailed study is needed to resolve this issue. Third, the monoenergetic imaging in this study was used to postprocess the dual-energy CT data; the technique may be limited to the patients who underwent dual-source, DECT acquisition. In conclusion, this study shows that the monoenergetic imaging of DECT improves image quality of CT in the patients with implanted metal orthopedic devices compared with the average weighted 120 kVp image. The optimal image quality with the lowest metal artifact can be extracted at 130 keV for the total, external, and internal implanted metal orthopedic devices compared with the average weighted 120 kVp images.
MONOENERGETIC IMAGING OF DUAL-ENERGY CT
4.
5.
6.
7.
8.
9.
10.
11.
12.
13. 14.
15.
16.
17.
REFERENCES 1. Watzke O, Kalender WA. A pragmatic approach to metal artifact reduction in CT: merging of metal artifact reduced images. Eur Radiol 2004; 14: 849–856. 2. Kalender WA, Hebel R, Ebersberger J. Reduction of CT artifacts caused by metallic implants. Radiology 1987; 164:576–577. 3. Yazdia M, Gingras L, Beaulieu L. An adaptive approach to metal artifact reduction in helical computed tomography for radiation therapy treatment
18.
19.
planning: experimental and clinical studies. Int J Radiat Oncol Biol Phys 2005; 62:1224–1231. Liu PT, Pavlicek WP, Peter MB, et al. Metal artifact reduction image reconstruction algorithm for CT of implanted metal orthopedic devices: a work in progress. Skeletal Radiol 2009; 38:797–802. Yu L, Li H, Mueller J, et al. Metal artifact reduction from reformatted projections for hip prostheses in multislice helical computed tomography techniques and initial clinical results. Invest Radiol 2009; 44: 691–696. Prell D, Kyriakou Y, Kachelrieß M, et al. Reducing metal artifacts in computed tomography caused by hip endoprostheses using a physicsbased approach. Invest Radiol 2010; 45:747–754. Prell D, Kyriakou Y, Struffert T, et al. Metal artifact reduction for clipping and coiling in interventional c-arm CT. AJNR Am J Neuroradiol 2010; 31: 634–639. Zhang Y, Zhang L, Zhu XR, et al. Reducing metal artifacts in con-beam CT images by preprocessing projection data. Int J Radiat Oncol Biol Phys 2007; 67:924–932. Lee MJ, Kim S, Lee SA, et al. Overcoming artifacts from metallic orthopedic implants at high-field-strength MR imaging and multi-detector CT. Radiographics 2007; 27:791–803. Link TM, Berning W, Scherf S, et al. CT of metal implants: reduction of artifacts using an extended CT scale technique. J Comput Assist Tomogr 2000; 24:165–172. Hemmingsson A, Jung B, Ytterbergh C. Dual energy computed tomography: simulated monoenergetic and material-selective imaging. J Comput Assist Tomogr 1986; 10:490–499. Fletcher JG, Takahashi N, Hartman R, et al. Dual-energy and dual-source CT: is there a role in the abdomen and pelvis. Radiol Clin N Am 2009; 47: 41–57. Johnson TR, Krauss B, Sedlmair M, et al. Material differentiation by dual energy CT: initial experience. Eur Radiol 2007; 17:1510–1517. Zhang LJ, Zhao YE, Wu SY, et al. Pulmonary embolism detection with dual-energy CT: experimental study of dual-source CT in rabbits. Radiology 2009; 252:61–70. Zhang LJ, Chai X, Wu SY, et al. Detection of pulmonary embolism by dual energy CT: correlation with perfusion scintigraphy and pathohistological findings in rabbits. Eur Radiol 2009; 19:2844–2854. Zhang LJ, Wu SY, Niu JB, et al. Dual energy CT angiography for the evaluation of intracranial aneurysms: image quality, radiation dose, and comparison with 3D rotational digital subtraction angiography. AJR Am J Roentgenol 2010; 194:23–30. Zhang LJ, Peng J, Wu SY, et al. Dual source dual energy CT of acute myocardial infraction: correlation with histopathological findings in a canine model. Invest Radiol 2010; 45:290–297. Thomas C, Heuschmid M, Schilling D, et al. Urinary calculi composed of uric acid, cystine, and mineral salts: differentiation with dual-energy CT at a radiation dose comparable to that of intravenous pyelography. Radiology 2010; 257:402–409. Bamberg F, Dierks A, Nikolaou K, et al. Metal artifact reduction by dual energy computed tomography using monoenergetic extrapolation. Eur Radiol 2011; 21:1424–1429.
1257