Radiation Dose Concerns in Cardiac Computed Tomography
Computed tomography (CT) utilisation for general medical imaging and for dedicated cardiac indications has come under a great deal of scrutiny in the past few years.1–3 Concern has arisen due to both the increased utilisation of CT, with more than 70 million scans performed in the US in 2007, and the increasing radiation dose. A recent report on ionising radiation exposure from medical imaging to the American population estimated that the collective radiation dose received increased by 700% between 1980 and 2006.4 CT has experienced an annual growth rate of >10% and accounted for approximately 50% of the total collective dose in 2006.
With the recent advances in CT gantry technology, including slip rings, multisegmented detector arrays and sub-second gantry rotation, there have been significant improvements in image acquisition time, enabling coronary CT angiography (CCTA).5 In 2006, cardiac CT was estimated to have resulted in 1.5% of the collective CT dose, but with significant further improvements in scanner technology and greater adoption of cardiac CT in clinical practice this number will almost certainly increase.
Cardiac CT provides excellent cross-sectional anatomical detail of the coronary arteries, the heart and the surrounding structures. It is unique in its ability to non-invasively image coronary atherosclerosis,6,7 as well as to provide a moderate descriptor of stenosis.8 However, this exceptional anatomical detail and spatial resolution has come at the price of a high historical radiation dose. The international Prospective Multicenter Study On RadiaTion Dose Estimates Of Cardiac CT AngIOgraphy I (PROTECTION I) trial surveyed 50 coronary CT angiography sites worldwide and reported median doses of 12mSv per exam.9 The majority of scans were performed with retrospective gating; use of prospective gating was limited to 6% of studies as the data were collected prior to widespread clinical release of this technique.
Radiation risk is a controversial topic in CT. The controversy stems from the fact that a direct link between radiation from medical imaging and development of solid-organ cancers has never been established in the <20mSv effective dose range encountered in cardiac CT. Despite the lack of direct evidence, most physicians and imaging societies advocate the adoption of the linear no-threshold hypothesis, which states that there is no threshold below which radiation cannot cause malignancy and that the risk of malignancy increases in a linear fashion with higher doses of radiation.
This hypothesis suggests that it is reasonable to linearly extrapolate the risk of malignancy induction from higher doses to the risk from lower dose exposures. It is this default position that requires that physicians take a conservative approach and adhere to the ‘as low as reasonably allowed’ (ALARA) principle, all the while maintaining diagnostic efficacy.
Imaging physicians must balance the benefits of the information obtained from CCTA with the potential risk of the radiation used to generate the images. If there is no proven benefit to performing an exam, regardless of the radiation dose the risk is too high and it should not be performed. Lowering the radiation dose of an appropriately ordered CT to a point where the scan loses diagnostic capability also does not ultimately benefit the patient. Likewise, use of greater amounts of radiation than needed likely will not add significantly to the diagnostic ability of the study.
Techniques used for dose reduction in cardiac CT can be grouped into optimised selection of scanning parameters and use of newer dose-reduction technology. Use of optimal user-selectable options includes methods such as weight- or body mass index (BMI)-dependent tube current, use of lower tube voltage,10,11 minimisation of the scan length (z-axis) and reduction in the number of phases or series performed. Newer scanning technology includes use of electrocardiogram (ECG)- dependent tube current modulation,12 prospective ECG gating,13–17 high-pitched spiral acquisition18–22 and use of noise reduction algorithms such as iterative reconstruction.23–25 Depending on the CT scan platform being used and the software version available, one may have many different tools available to lower radiation dose. Many of the techniques described below can and often are used in combination for additional dose reduction.
Measuring Radiation Dose
Radiation dose is proportional to the tube current, the exposure time and the square of tube voltage and is inversely proportional to the pitch for helical acquisition. Estimated radiation doses for CCTA examinations can be expressed in numerous terms. Conventional units of radiation dose (rads and rems) have been replaced with SI units of Grays (Gy) and Sieverts (Sv). Special measures for radiation emitted from CT have been developed: the volume computed tomography dose index (CTDIvol [in Gy]), dose length product (DLP [in mGy x cm]) and effective dose (E [in mSv]).
The CTDIvol averages radiation dose over x, y and z directions. This is used to express the average dose delivered to the scan volume (3D CT slice) for a specific examination. The DLP is defined as the CTDIvol times scan length and is an indicator of the integrated radiation dose of an entire CT examination. The CTDIvol and DLP are now reported by most CT systems in current use. E is determined from dose to individual organs and the associated relative radiation risk assigned to each organ. The estimated effective dose for a patient is obtained by multiplying DLP by a conversion factor, k (in mSv x mGy-1 x cm-1), which varies depending on the body region that is imaged. These normalised effective dose coefficients are determined by the radiation sensitivity of the body region scanned based on exposed organ radiosensitivities. In chest CT, the accepted standard for effective dose quantification is 0.017, which, when multiplied by the DLP in mGy x cm, allows for the calculation of the study effective dose in mSv. It should be noted that while 0.017 is the most commonly used conversion factor for chest imaging, the current recommended conversion coefficient for CCTA is 0.014mSv mGy-1cm-1.26
Techniques for Dose Reduction
Tube Current Optimisation
Current cardiac-capable CT scanners have significantly greater tube power than earlier machines. Cardiac protocols require greater tube current delivered in a significantly shorter period of time than is routinely used for other CT protocols. This power is useful for rapid cardiac scanning of individuals with high BMI. However, if protocols are not individually adjusted, the result may be needlessly high radiation doses.
Anatomy-adapted tube-current modulation, commonly used in non-cardiac applications, is not fully compatible with the ECG-dependent dose-modulation technique currently used in CCTA. The appropriate tube current must be manually selected for each case. The current is tailored according to the patient’s BMI, chest circumference and estimated muscle and breast mass. Reliance on a ‘standard CT protocol’ without altering mA may lead to excessive radiation doses for thin patients and potentially poor image quality for high-BMI patients. Additionally, a 5cm difference in thoracic diameter corresponds to a factor of two or more in the dose required to maintain similar image quality.27 Use of weight-adapted mA can reduce the dose by 18–26%, while constant image noise is achieved and image quality preserved.28
In the PROTOCOL study, LaBounty et al. prospectively evaluated 449 patients undergoing 64-detector CCTA at three centres and compared them pre- (n=247) versus post-initiation (n=202) of a standardised BMI-and heart-rate-based protocol that incorporated multiple dosereduction strategies, including gating technique, tube voltage, current, padding duration and scan length.29 In multivariate analysis, a 20% reduction in radiation dose was associated with every 100mA reduction in tube current.
Tube Voltage Optimisation
As the radiation dose varies with the square of the kilovoltage, small reductions in tube voltage enable a marked reduction in effective dose (see Figure 1).10,30–32 Abada et al. reduced the kV in CCTA from the traditional 120 to 80kV in low-BMI patients undergoing CCTA and found up to an 88% dose reduction.10 Use of 100kV in the PROTECTION I study resulted in a 53% reduction in dose; image quality was maintained.33 Lowering the kV in CCTA exams also increases noise. The prospective randomised PROTECTION II study randomised 400 patients with a bodyweight <90kg to either standard 120kV or 100kV scan protocol.34 Radiation dose was significantly lower in the 100kV group, with a 31% dose reduction without significant difference in qualitative or quantitative image assessment.
Decreasing voltage to 80 or 100kV has the added benefit of also increasing opacification of vessels due to an increase in the photoelectric effect and a decrease in Compton scattering. Intraluminal CT attenuation (HU) of the coronary arteries is significantly higher at 100kV compared with 120kV.35 Due to this increase in opacification, the iodine dose can be reduced safely.31,36
Scan Coverage Optimisation
Radiation dose is directly proportional to z-axis coverage. Accurate prescription of anatomical coverage is important to minimise dose. If the scan length is too long, unnecessary radiation is delivered to the upper chest and abdominal organs. If the length is too short, part of the coronary tree may be excluded. When minimum z-axis coverage is prescribed, there is a risk of excluding anatomy if the patient does not perform an identical breath-hold to the one used for planning purposes. Teaching the patient to perform a breath-hold in the same manner every time will help avoid truncation of anatomy because of irregular breath-holding.
Minimisation of the z-axis is important. In the PROTOCOL study, multivariate analysis showed a 5% reduction in effective dose for every 1cm of reduced z-axis scan length.29,37 To minimise z-axis coverage but ensure adequate anatomy is imaged, we prescribe the z-axis from review of the calcium-scoring scan, if performed, or we perform a very-low-dose axial scout. CCTA scanning starts 20mm above the left main origin and concludes 10mm below the cardiac apex. Other sites prefer to plan coverage from the scout topogram to prescribe a volume beginning at the bifurcation of the trachea. However, use of the scout film has been shown to be inaccurate for determining the precise origin of the coronary arteries.38
Field of View
Limitation of the field of view (xy plane) helps minimise radiation and potentially improves quality. The size of the imaging voxel is reduced as the field of view is decreased, improving spatial resolution. Use of a bow-tie filter allows for reduced radiation exposure by limiting the scatter towards the detectors. These filters are optimised for patient size. Most cardiac studies can be imaged with a small bow-tie filter, which is most efficient at reducing dose. Budoff et al. reported a 40% dose reduction by simply using a small bow-tie filter rather than larger sized filters.39 Routine use of small bow-tie filters and a small field of view is prudent to minimise the dose of cardiac CT studies.
Electrocardiogram-controlled Tube Current Modulation
Scan data used for diagnosis are most commonly from the diastolic phase at a time of relative cardiac quiescence. As a result, a high tube current is required during the diastolic phase only and a lower tube current is acceptable during the remainder of the cardiac cycle. Reduction of the tube current during the less critical portions of the cardiac cycle is achieved using ECG-gated tube current modulation. ECG-gated tube current modulation reduces radiation exposure without decreasing image quality. Using this technique, full tube current is on during the most optimal phases of the cardiac cycle and is then reduced during the remaining phases (see Figure 2).
ECG-gated tube current modulation algorithms allow for a varied length of full tube current plateau depending on the patient’s heart rate. Depending on heart rate, ECG-gated tube current modulation can reduce the overall effective radiation dose by 20–48%.10,12,30 The effects of dose reduction are more pronounced for lower heart rates. In the PROTECTION I study, a 25% dose reduction was noted when ECG-gated tube current modulation was used, which was the case in 73% of studies.9
Benefits of ECG dose modulation are dependent on patient heart rate. At higher heart rates, there is less time to rapidly change the tube current, and a less substantial reduction in effective dose will be obtained. In addition, since reconstruction of the coronary arteries in late systole is often of use at higher heart rates, a longer peak current plateau may be helpful and, at times, necessary; this unfortunately drives the dose higher. The impact of longer maximum tube current plateau scanning is so great that some centres elect to either extend the time of full tube current or forgo its use altogether, increasing the relative dose. In patients with irregular heart rates, ECG-gated tube current modulation may inadvertently reduce tube current during the optimal imaging phase, yielding reduced image quality and diagnostic utility.30 Some newer versions can detect arrhythmias and automatically maintain full tube current for the arrhythmic cardiac cycles.
Prospective ECG triggering, also referred to as ‘step-and-shoot’ or ‘sequential’ scanning, has shown the largest effective dose reduction compared with the other methods discussed here.13-15,17,40–43 Using this technique and a scanner with a wide detector, the table is stationary during image acquisition and moves to the next location for another scan that is initiated by the subsequent cardiac cycle. There is very little overlap between the scans, resulting in a dose reduction. Complete coverage of the heart can be performed in as little as a single heartbeat for 256- and 320-row multidetector-row CT (MDCT) scanners,44–47 two to three heartbeats with 128-detector dual-source CT (DSCT)48,49 or five to eight heartbeats on a 64-row system.13
Prospective ECG triggering enables acquisition of CCTA at doses generally in the 1–4mSv range (see Figure 3).13–15,17,40–43 We first reported a mean effective dose for prospectively triggered exams of 2.8mSv – 83% less than with the retrospective helical technique (mean 18.4mSv).13 Despite the reduced dose, the image quality score was significantly greater for images obtained with a prospectively triggered technique. Many other studies have since evaluated the use of prospective gating for CCTA. In the PROTECTION I study, prospective ECG triggering led to a 71% reduction in effective radiation dose but was used in only 6% of studies.9 The 6% figure reflects the fact that the PROTECTION I data were accumulated prior to widespread release of prospective triggering protocols.
Diagnostic accuracy of prospectively triggered CCTA compares well with conventional coronary angiography (CA). Scheffel et al. reported overall patient-based sensitivity, specificity and positive (PPV) and negative predictive value (NPV) for the diagnosis of significant stenoses of 100, 93, 94 and 100%, respectively.43 Stolzman et al. reported overall sensitivity, specificity, PPV and NPV of 98, 99, 95 and 100%, respectively.50 Dewey et al. recently evaluated 320-row MDCT before same-day CCA and reported per-patient sensitivity and specificity of 100 and 94%, respectively.46
There are limitations to prospective ECG triggering. There is no capability to evaluate cardiac function because data are acquired only during a limited portion of the cardiac cycle. Most studies performed with prospective triggering have used upper heart rate criteria of between 65 and 70bpm because image quality degrades at higher heart rates. Husmann et al. reported that 98.9% of the coronary segments were clinically assessable for heart rates less than 63bpm; however, only 85% were assessable at higher heart rates.41 Other studies have reported a similar inverse correlation between image quality or segment assessability and heart rate.15,51,52 Heart rate variability has also been independently correlated with overall image quality for prospectively triggered CCTA.15,51,52 Newer techniques and faster CT systems may be helpful in raising the upper heart rate limitation. Using prospectively triggered DSCT, Xu et al. evaluated patients with a heart rate of 70–110bpm and found coronary evaluability of 99.7%, similar to 98.7% with retrospective gating.53
Prospective triggering at most centres relies on the use of betablockers to lower the heart rate. In our clinical practice we use prospective triggering for all cases as long as the heart rate is less than 70bpm, heart rate variability is less than 10bpm and cardiac function is not required. In an analysis of 2,124 consecutive clinical exams, approximately 50% of patients initially failed one or more of the listed criteria.42 However, with careful heart rate control, prospective triggering was eventually used for 92.1% of coronary CTAs and 83.2% of CTAs following coronary artery bypass graft (CABG).
High-pitch Spiral Prospectively Triggered Dual-source Computed Tomography
With the recent introduction of second-generation DSCT systems, a new fast-table-speed/high-pitch-scan mode has been developed.18–22 Achenbach et al. recently evaluated this scan mode in 37 consecutive patients with bodyweight <100kg and low heart rates (<60bpm).19 They reported an estimated effective radiation dose of 0.94±0.06mSv without degradation of image quality.
This technique utilises a very high pitch factor of 3.2 and two large (64-slice) detectors with scan initiation at 60% of the R–R interval. This high helical pitch of 3.2 was previously not possible with single-source scanners. With single-source techniques, the spiral or helical pitch (table movement per rotation divided by the collimated beam width) is limited to a maximum of 1.5 to ensure gap-less coverage in the z-axis. A DSCT has two X-ray tubes and two detectors arranged at a 90º angle, allowing a high temporal resolution and high-pitch data acquisition. In the high-pitch scan mode, the second tube detector of a DSCT system is used to essentially fill the data gaps, enabling the helical pitch factor of 3 to be used. Due to the short sub-second radiation exposure and little overlap between spiral acquisitions, considerable reduction in radiation dose has been found.
This technique is not without limitations. Similar to prospective triggering, no functional data are available, and the technique is susceptible to artifacts due to ectopy. Use of the high-pitch spiral technique has been limited to heart rates <60bpm, and it has been primarily used with non-obese (<100kg) patients. Proof-of-concept trials were initially performed on the 64-row DSCT system; however, optimally the latest 128-detector row DSCT platform should be used. Nevertheless, early clinical results show good potential and the technique represents another method for performing low-dose CCTA.
Image reconstruction for CT has traditionally been performed using filtered back projection (FBP). FBP is fast and mathematically simple, requiring limited computational power, but has limitations that negatively affect image quality. Adaptive statistical iterative reconstruction (ASIR) is a new technique that may be used to lower the radiation exposure of all CT studies, including cardiac.23–25,54 ASIR uses iterative comparisons of each acquired image to a synthesised projection incorporating modelling of both system optics and system statistics. Compared with FBP, images reconstructed with ASIR have lower image noise. By using ASIR to reduce image noise, one can reduce tube current and maintain relative image noise, achieving lower radiation doses.
Hara et al. reported that use of ASIR was associated with 32–65% effective dose reductions compared with FBP without degradation of image quality for abdominal imaging.23 In CCTA, Leipsic et al. evaluated 62 patients using an ASIR-capable 64-detector scanner and a low-dose CCTA technique.55 In comparison with FBP (0% ASIR), increasing ASIR reduced image noise by up to 43%, but there was no effect on signal (see Figure 4). Reconstruction using 40 or 60% ASIR significantly improved image quality and the proportion of interpretable segments when using a low mA technique.
Use of ASIR was recently reported in a multicentre analysis in 1,202 consecutive patients in three arms: routine 64-detector CCTA with FBP (n=753), 64-detector CCTA with ASIR (n=247) and the latter following initiation of a protocol using 100kV (in patients with a BMI <30) and a BMI-based mA protocol (n=202).37 The mean effective radiation of the three groups was 3.8, 2.6 and 1.3mSv, respectively. Despite the overall radiation dose reduction, image interpretability and signal-to-noise ratio of post-protocol ASIR and FBP were similar.
There has been heightened concern of a risk from increasing radiation exposure from cardiac and general diagnostic CT imaging. This has led to a great deal of research on new CT techniques that are capable of imaging the heart at lower radiation doses. In clinical practice, optimised selection of user-defined parameters such as tube current and voltage, as well as use of new technologies such as prospective triggering, can lead to a substantial reduction in radiation dose without loss of diagnostic accuracy.