In recent years, CT is playing a more important role in the diagnosis of chest diseases, especially lung cancer screening. Meanwhile, the radiation exposure1 caused by applications of CT in modern medicine has significantly increased. Studies have shown that one X-ray chest film would produce a radiation dose of about 0.7 mSV, but the dose would be up to 5.1–11.1 mSv for each chest CT scan.2 The risks of iatrogenic radiation exposure along with CT examinations have been a topic of debate among medical professionals, but recent data suggest it may significantly increase the risk of adverse effects.3 Currently, the reduction of the radiation dose, while maintaining image quality, is a very hot research field.
There are many technologies for reducing radiation dose of CT scanners, such as reduced tube voltage, automated tube current modulation, and decreased scan length,4,5 but the dose reduction is limited because of the current reconstruction method of filtered back projection (FBP). A new method based on iterative reconstruction algorithms has been implemented in the last generation of CT equipment. With this technique, image data are corrected by comparing individual pixel value against the ideal value of a generated noise map that the noise model predicts. Through multiple repetitive reconstructions, it subsequently produces optimization of image quality.6 The adaptive statistical iterative reconstruction (ASIR) approach used by the Gemstone Spectral CT HD750 is a commercially available technology based on the choice of the percentage of iterative algorithm.7 ASIR is likely to greatly reduce noise while ensuring good image quality, so as to effectively reduce the patient radiation dose. In this area some researchers have conducted studies using models and clinical experiments.7 Several studies demonstrated the possibility of reducing patient dose by 25%-60% with iterative reconstruction versus FBP.8,9 It is already accepted that the ASIR technology can reduce the radiation dose of CT examinations. But the routine, standard-parameters for low-dose chest CT scanning have not been defined. In this study, 72 adult patients were examined by spectral CT with different combinations of scanning parameters. All of the images were analyzed in order to evaluate the radiation dose and image quality among different groups and to devise standard parameters for chest examination by Gemstone spectral CT HD750.
The ethics committee of the First Hospital of Jilin University had made a strict review of the research schemes and had approved our study; and Health Insurance Portability and Accountability Act compliant practices were used during this study. Participants’ written informed consent was obtained.
Over a 7-month period (November 2011 to May 2012), 72 adult patients (47 males and 25 females; age 25–95 years old, average 62.36 years with an average BMI of 23.56 kg/m2) were prospectively enrolled in this study. The patients were referred for a plain chest CT for regular check-ups or to identify some lesions suspected on chest radiography.
All CT scans were performed on an ASIR-capable multi-detector row CT machine (Gemstone Spectral CT Discovery HD 750, General Electric Healthcare, USA). Before scanning, all of the metals carried by the participants were removed to avoid artifacts, and we trained the participants in their breathing to cooperate with the scanning. The scan was obtained in the supine position during a single inspiratory breath hold. The scan range was from the level of the pulmonary apex to the diaphragm. Scanning parameters were as follows: tube voltage 120 KV; tube current was automatic mA technology from 650 mA to 50 mA; scanning pitch 0.984:1; scanning time 0.6 s/circle; standard reconstructive mode and slice thickness 5 mm. The noise index (NI) was set after scouting. Subjects were randomly divided into six groups. In each group, there was a different NI value: 13.0, 15.0, 17.0, 19.0, 21.0 and 23.0 HU, respectively. During reconstruction, the percentage of ASIR was controlled in the range of 30%-70% and the distance was 20%. So we gained a total of 18 image sequences.
Objective image quality evaluation
The mean CT number and standard deviation (SD) of the CT attenuation in the central descending aorta and the paraspinal muscle on both sides were measured at the level of the tracheal carina using a region of interest (ROI) of 95–100 mm2 in all of the reconstructive images. The signal-tonoise ratio (SNR) and contrast-to-noise ratio (CNR) were calculated using the following equations: SNR=CTaorta/ image noise; CNR=(CTaorta-CTmuscle)/image noise, where CTaorta represented the mean CT attenuation of the descending aorta, CTmuscle represented the mean CT attenuation of the paraspinal muscle on both sides, and image noise represented the SD of CT attenuation of the descending aorta.
Subjective image quality evaluation
The reconstructive images were displayed randomly on a high-resolution diagnostic monitor using our hospital's picture archiving and communication system (PACS) workstation. They were evaluated in a blinded fashion by two radiologists with more than five years of imaging diagnostic experience. The image details that should be evaluated were: Lung window (1500 HU window width/-700 HU window center); with attention to the show of lung texture, bronchial walls, vessels and bilateral hilus of the lung. Mediastinal window (340 HU window width/38 HU window center); with attention to the mediastinal structure, muscle profile and chest wall. Spinal window: with attention to the show of bone structure and profile. The overall quality: the perceived noise, image contrast, the hierarchies of different tissues, edge sharpness and the artifacts. Based on the above criteria, we set a five-point grading scale as follows: 1 point when below the standard and it can not meet all the requirements of diagnosis; 2 points when also not good enough to be used for diagnosis; 3 points when just about satisfying the diagnostic acceptability; 4 points for a good image; and 5 points for the best image.
Radiation dose measurements
The volume CT dose index (CTDI) and dose length product (DLP) values were displayed on the console monitor of the CT scanner. The effective dose (ED) was calculated to estimate the radiation dose by multiplying DLP values with a conversion factor of 0.014 mSv/mGy·cm for chest CT in adults.10
The data were analyzed using SPSS 19.0 (SPSS Inc., USA). CT values were compared using the Student's t-test, and SD, SNR, CNR, CTDI, DLP, and ED were compared using the Variance analysis. Among the six groups with different NI, we used Tamhane's T2 test to assess the objective data because of the heterogeneity of the variance. The consistency of the different radiologists’ scores was assessed by a Kappa test. Kappa ≥0.75 means the consistency is very good, 0.75 >Kappa ≥0.4 means the consistency is common, Kappa ≤0.4 means the consistency is bad. The difference was judged to be statistically significant when a P value was less than 0.05.
Analysis of objective image quality
The CT values of the descending aorta with the 18 image settings had no significant difference (Table 1).
Under the same NI, the SD of the CT attenuation in the central descending aorta was diminished as the ASIR increased (Table 2), and there was a significant difference among different ASIR (when NI 13.0 P <0.0001; NI 15.0, P=0.033; NI 17.0, P =0.001; NI 19.0, P=0.034; NI 21.0, P=0.002, and NI 23.0, P <0.0001). Under the same ASIR, the SD was increased as the NI got bigger (Table 2). The results of inter-group comparison suggested there was a significant difference of SD between the NI 13.0 group and the NI 15.0, 17.0, and 23.0 groups.
For the SNR and CNR there was a trend towards gradual reduction as the NI increased (Table 3). There was no significant difference among the groups for SNR and CNR when the NI was 15.0, 17.0, or 19.0.
In this study, when the NI was up to 23.0, there were a large number of artifacts. The pixels became over-smooth and the image contrast decreased significantly (Figures 1 and 2), so that the NI could not be increased any more.
Analysis of subjective image quality
As SD was related to the noise level, the smaller the SD, the better the image quality. Based on the trend shown in Table 2, we chose four groups (NI 13.0, 15.0, 17.0, and 19.0) to evaluate and score the image quality. The results are shown in Figure 3. The consistency of the two doctors’ scores were common with a kappa=0.657.
The CTDI, DLP, and ED were decreased as NI increased (Table 4). For CTDI and DLP, there was a significant difference between the NI 13.0 group and the NI 17.0, 19.0, 21.0, and 23.0 groups. There was also a difference between the NI 23.0 group and the NI 15.0, 17.0, 19.0 groups in CTDI and DLP.
Based on the early domestic and foreign research in lowdose spectral CT, we think the conclusion that ASIR could reduce the radiation dose is definite. The noise is decreasing as the ASIR increases, and there is a strong positive correlation between them.11–14 How best to balance the relationship of NI, ASIR and image quality, while setting the best scanning parameters of NI and ASIR has not been discussed in depth. This study was projected to solve the problem and produce a reference and guidance for clinical practice.
The technology of ASIR and the best level of ASIR
One of the most important factors that influence CT radiation dose and image quality is the image reconstruction algorithm. Traditionally, CT mainly uses the filtered back projection (FBP), with an advantage of fast reconstruction while the disadvantage is that it is more sensitive to noise, since FBP does not take into account certain system hardware details (such as actual focal spot, detector sizes, and location) and system noise (such as photon statistics and electronic noise).15 Compared with FBP, ASIR is a new technology of statistical iterative reconstruction, using a statistical model to reduce image noise and improve the low contrast detestability of image. It has a huge potential for improving image quality14,16,17 and lowing radiation dose through reducing the noise. That is to say, the image quality will be improved at low doses and, if the noise remains the same, the dose will be lower than with common scanning.
In theory, the higher the ASIR weight, the larger the capability of eliminating noise. But when ASIR reaches a point, 70% for example, the image contrast would become too low and the image quality would be too poor to satisfy imaging diagnosis.18 In practical work, both the actual effects of image quality and radiation dose should be considered. A choice of 40% ASIR implies that 40% of the ASIR image was blended with the FBP image. A 40% ASIR is considered to be a reference value in some studies19 and it is also the recommended value by the manufacturer. Through the analysis of different settings from 0 to 100% ASIR weight, the ranges of 0–30% and 70%-100% ASIR produced an unacceptable image quality and they were not used. We chose 30%, 50% and 70% ASIR as our research parameters. Objective results showed the SD was decreased as the ASIR% increased and the difference was statistically significant, the degree of reduction was 23%- 28% in the six groups, and the SNR and CNR increased along with ASIR percentage rising up. So a bigger ASIR is more suitable for a standard parameter. The analysis of subjective image quality showed the score became smaller with an increase of ASIR (Figure 3). In theory, a smaller SD means a less noise, the image quality should be better, and the results of SNR and CNR also demonstrate a higher image quality when ASIR increases. However, our results are the opposite of this conclusion. We speculate that this may have some connection with the process of CT reconstruction. When ASIR is at a high degree, the machine will reconstruct just the same as at a low scanning dose. So its inherent noise is much greater than at a low ASIR weight.13 Another factor is that the noise structure will change when reconstructing at a much high ASIR and the image will lack fidelity; therefore, our doctors gave low scores to the images reconstructed by high ASIR. Through statistical analysis of these six groups, we found that when NI was fixed, there was no significant difference between the scores from 30% ASIR and 50% ASIR. But the difference was significant between 30% and 70% ASIR, and also between 50% and 70% ASIR. In addition, the SD values were significantly different between 30% and 50% ASIR, therefore the best ASIR is 50%.
The best choice of NI
There were six groups in this study based on different NI. The main observed indexes were the scores of image quality and radiation dose. When NI was 23.0 HU, the image quality was too poor to satisfy clinical diagnosis. That is to say, when the NI increased to a critical level, the data from the images started to become unreliable. We could not improve the NI indefinitely to reduce the radiation dose using ASIR technology.
When the ASIR was constant, the greater the NI and SD, the lower the radiation dose (Tables 2, 4). The Student's t test results between NI 15.0 and 17.0 showed that the P values were P=0.173 and P=0.126 respectively. And the P values for CTDI and DLP between NI 17.0 and 19.0 were not statistically significant, P=0.0957 and P=0.975. We think this may be because the sample size was somewhat small. Based on subjective aspects, we chose four groups to evaluate through a preliminary analysis of 18 image settings with NI values of 13.0, 15.0, 17.0, and 19.0. The curves of the two radiologists’ scores were all declining as the NI improved (Figure 3). The score of NI 17.0 was slightly higher than that of NI 15.0, but this was not a statistically significant difference. The P values at 30%, 50%, and 70% were 0.948, 1.256, and 0.975, while between NI 17.0 and NI 19.0, the difference was very clear with a P value < 0.05. Therefore, the best NI should be 17.0. In this case, the average ED was about 1.94 mSV, it decreased by about 36% when compared to NI 13.0, and 27% compared to NI 15.0.
In conclusion, the best parameters in adult chest scanning by spectral CT should be as follows: NI 17.0 and ASIR 50%. The ED would be smaller than that of a routinedose CT and this is in the range of the safe dose provided by the International Commission Radiological Protection (ICRP).20 It can also ensure the image quality for diagnosis. The main disadvantages of this research are the size of sample is not large enough because of time constrains, and the classification of ASIR weight should also be more sophisticated to get more valuable data. Further detailed research in the future is expected.
1. Nevzat K, Maeit A. Low-dose CT: practices and strategies of radiologists in university hospitals. Diagn Interv Radiol 2006; 12: 3-8.
2. Fujii K, Aoyama T, Koyama S, Kawaura C. Comparative evaluation of organ and effective doses for pediatric patients with those for adults in chest
and abdominal CT examinations. Br J Radiol 2007; 80: 657-667.
3. Brenner DJ, Elliston CD. Estimated radiation risks potentially associated with full-body CT screening. Radiology 2004; 232: 735-738.
4. Li Q, Yu H, Zhang L, Fan L, Liu SY. Combining low tube voltage and iterative reconstruction for contrast-enhanced CT imaging of the chest
-initial clinical experience. Clin Radiol 2013; 68: 249-253.
5. Campbell J, Kalra MK, Rizzo S, Maher MM, Shepard JA. Scanning beyond anatomic limits of the thorax in chest
CT: findings, radiation dose, and automatic tube current modulation. AJR Am J Roentgenol 2005; 185: 1525-1530.
6. Silva AC, Lawder HJ, Hara A, Kujak J, Pavlicek W. Innovations in CT dose reduction strategy: application of the adaptive statistical iterative reconstruction
algorithm. AJR Am J Roentgenol 2010; 194: 191-199.
7. Rampado O, Bossi L, Garabello D, Davini O, Ropolo R. Characterization of a computed tomography
iterative reconstruction algorithm by image quality
evaluations with an anthropomorphic phantom. Eur J Radiol 2012; 81: 3172-3177.
8. Hu XH, Ding XF, Wu RZ, Zhang MM. Radiation dose of nonenhanced chest
CT can be reduced 40% by using iterative reconstruction in image space. Clin Radiol 2011; 66: 1023-1029.
9. Pontana F, Duhamel A, Pagniez J, Flohr T, Faivre JB, Hachulla AL, et al. Chest computed tomography
using iterative reconstruction vs filtered back projection (Part 2): image quality
of low-dose CT examinations in 80 patients. Eur Radiol 2011; 21: 636-643.
10. Zhang XH, Yang L, Wu J, Ju HY, Zhang F, He B, et al. Diagnostic accuracy and its affecting factors of dual-source CT for assessment of coronary stents patency and in-stent restenosis. Chin Med J 2012; 125: 1936-1940.
11. Hara AK, Paden RG, Silva AC, Kujak JL, Lawder HJ, Pavlicek W. Iterative reconstruction technique for reducing body radiation dose at CT: feasibility study. AJR Am J Roentgenol 2009; 193: 764- 771.
12. Vorona GA, Ceschin RC, Clayton BL, Sutcavage T, Tadros SS, Panigrahy A. Reducing abdominal CT radiation dose with the adaptive statistical iterative reconstruction
technique in children: a feasibility study. Pediatr Radiol 2011; 41: 1174-1182.
13. Rapalino O, Kamalian S, Payabvash S, Souza LC, Zhang D, Mukta J, et al. Cranial CT with adaptive atatistical iterative reconstruction: improved image quality
with concomitant radiation dose reduction. AJNR Am J Neuroradiol 2012; 33: 609-615.
14. Marin D, Nelson RC, Schindera ST, Richard S, Youngblood RS, Yoshizumi TT, et al. Low-tube-voltage, high-tube current multidetector abdominal CT: improved image quality
and decreased radiation dose with adaptive statistical iterative reconstruction
algorithm—initial clinical experience. Radiology 2010; 254: 145-153.
15. Prakash P, Kalra MK, Digumarthy SR, Hsieh J, Pien H, Singh S, et al. Radiation dose reduction with chest computed tomography
using adaptive statistical iterative reconstruction
technique: initial experience. J Comput Assist Tomogr 2010; 34: 40-45.
16. Thibault JB, Sauer KD, Bouman CA, Hsieh J. A threedimensional statistical approach to improved image quality
for multi-slice helical CT. Med Phys 2007; 34: 4526-4544.
17. Prakash P, Kalra MK, Kambadakone AK, Pien H, Hsieh J, Blake MA, et al. Reducing abdominal CT radiation dose with adaptive statistical iterative reconstruction
technique. Invest Radiol 2010; 45: 202-210.
18. Leipsic J, Labounty TM, Heilbron B, Min JK, Mancini GB, Lin FY, et al. Adaptive Statistical Iterative Reconstruction
: Assessment of Image Noise and Image Quality
in Coronary CT Angiography. AJR Am J Roentgenol 2010; 195: 649-654.
19. Flicek KT, Hara AK, Silva AC, Wu Q, Peter MB, Johnson CD. Reducing the radiation dose for CT colonography using adaptive statistical iterative reconstruction
: a pilot study. AJR Am J Roentgenol 2010; 195: 126-131.
20. Hall EJ. Is there a place for quantitative risk assessment? J Radiol Prot 2009; 29: A171-184.
Keywords:© 2014 Chinese Medical Association
chest; computed tomography; image quality; noise index; adaptive statistical iterative reconstruction