In our study, we scanned the Catphan CT phantoms using a conventional tube voltage and decreasing tube currents, measured noise and CNR of each dose group reconstructed using both FBP and IRT approaches, and identified the levels of IRT in each low dose group that provided IQ that was equivalent to the routine-dose FBP reconstruction. From the comparisons of IQ (both subjective and objective) between the routine-dose FBP group and each low- dose IRT group, the protocol of 120 kV, 84 mAs and L5 reconstruction was determined as the protocol that provided an optimal balance of IQ and radiation dose reductions. This protocol was also performed in patients with normal body habitus prospectively enrolled in a clinical study to further confirm the feasibility of this low-dose protocol.
The IRT used in this study (iDose4
) is a fourth generation algorithm designed to reduce image artifacts and noise while maintaining the structural/anatomical information 
. The use of photon statistics in the projection domain helps to iteratively reduce noise and preserve edges, and use of noise/structure models in the image domain to further reduce image noise.
The levels of IRT used in this study are designed to provide a noise reduction factor to compensate for any noise increases with the reduction in the tube output. Our phantom results showed a linear decrease in image noise with increasing levels of IRT; our experience was similar to prior work 
achieving noise reduction levels of 15%–45% as the levels of the IRT was increased from L2 to L6 for a given tube output. Thus, when the tube output is reduced by 40–70%, the corresponding IRT levels can be employed to result in objective image quality which is equivalent to or better than the FBP images from the routine-dose scans (i.e., Group A). In our phantom study, the noise values in the Groups B (125 mAs) and C (105 mAs) were actually lower than that those measured in Group A, and there were no significant differences in actual noise values between Groups D & E and Group A. Lowered image noise resulted in increased CNR in Groups B and C compared to Group A; no significant differences in CNR were observed between Groups D & E (59.3±5.3 and 59.5±2.3 respectively) and Group A (56.2±2.8)(P
There were no significant differences in MTF value and curve shape between Group A and Groups B–E. This suggests that the IRT decreases image noise, while not negatively impacting spatial resolution. Retaining a relatively high spatial resolution is of importance for displaying coronary arteries.
The subjective evaluation scores of phantoms were similar between each low-dose iterative reconstruction group (Groups B–E) and Group A. No graininess, streak artifacts or blotchy/plastic artifacts were observed in each group. This advantage is superior to early iterative reconstruction approaches 
. This can be attributed to the IRT used in this study which reduces noise without altering the image noise power spectrum and prevents artifacts to preserve the natural appearance of the image.
The subjective scores in Groups B and C were superior to those of Group A. Artifact, nodule conspicuity, minimal detectable nodule size and overall IQ in Group D were equivalent to those in the Group A. The nodule conspicuity and minimal detectable nodule size in the Group E were slightly lower than those in the Group A. These results may be related to the slightly decreased spatial resolution 
. Our phantom experimental results showed that according to the “ALARA” principle, a tube voltage of 120 kVp, 84 mAs tube current x-ray on time product supplemented by an iterative reconstruction technique of an appropriate level (L5) is the optimal combination that provides IQ comparable to the FBP reconstructions of routine-dose scans.
Recently, some studies investigating low radiation dose coronary CTA used a decreased tube voltage (80 kVp and 100 kVp) and maintained high IQ and CNR in a select group of patients with a normal body mass index 
. This could be attributed to the selected energy levels being closer to the K edge of iodine, thereby increasing the contrast enhancement and thus potentially enabling a reduction of volume of contrast agent used.
However, the use of low tube voltage could also result in a larger proportion of dose absorbed by the body, potentially causing a higher degree of beam hardening effects. Since reducing the tube output (current) instead of the tube voltage only changes the effective energy and affects the x-ray penetration to a lesser extent, we used a conventional tube voltage (120 kVp) and low tube currents in this study.
The prospectively ECG-triggered coronary CTA is currently the main method to decrease the radiation dose in coronary CTA. The 256-slice multi-detector computed tomography system can achieve an axial scan range of 8 cm and allow scan of the entire heart within 3 cardiac cycles, with a rotation speed of 0.27 sec, which could relax the heart rate requirements for coronary CTA using prospective gating (i.e., increases the threshold of the upper HR limit to 75 bpm) while at the same time enabling coronary CTA imaging at well under the average background radiation levels 
Extending our findings from the phantom scans, we have shown that the iterative reconstruction technique used in this study (iDose4), combined with prospective ECG gating in 256-slice MDCT can significantly decrease radiation dose in patient scans without negatively impacting IQ. Results from this study showed that there were no significant differences in subjective and objective IQ between Group 2 (reduced dose patient scans) and Group 1 (routine dose). The CTDI and ED in the Group 2 were decreased by 63% and 62% respectively compared to Group 1 (p<0.05). The effective radiation dose in the Group 2 reached 1.21±0.14 mSv. The extent of radiation dose savings is similar to recent findings 
. The dual benefits of maintaining (or even improving IQ) at low radiation dose could make the use of IRT in virtually all patient cohorts, especially in those sub-groups that are sensitive to radiation dose (younger population like infants and pediatrics) and those who require re-examination for follow-up of coronary plaque progression, percutaneous transcoronary angioplasty, coronary artery bypass grafting (CABG), etc.
The limitations of this study are as follows: (1) Our sample size was small and we did not adjust the protocol according to the BMI/body weight of the individual patients (the BMI ranged from 23 to 29 in our study). (2) We only focused on the overall image quality and did not investigate the effect of the IRT on coronary artery plaque and the diagnostic accuracy by comparing with the respective gold standards. (3) We did not evaluate the reconstruction time of the IRT compared to the conventional FBP, which may influence its wide clinical application; but the IRT used in this study had a reconstruction time of about 20 images per second making it clinically practical.
In conclusion, our results of phantoms and the subjects with normal body weights confirmed that the iterative reconstruction technique (iDose4) used in prospectively ECG-triggered 256-slice coronary CTA can greatly decrease image noise and improve image quality, while at the same time providing radiation dose reductions of up to 63%.