The overall increase in patient irradiation caused by the growing use of spiral- and multidetector-row computed tomography (MDCT) is of particular relevance for thoracic imaging. The number of clinical indications for thoracic CT has steadily increased, and CT has become a first-line imaging tool for diseases previously imaged with chest radiography, ventilation/perfusion scintigraphy and pulmonary angiography.1 Moreover, the use of CT for screening purposes has increased the number of CT examinations performed in clinically asymptomatic patients.2 Finally, the relatively higher number of CT examinations performed in younger patients increases cumulative radiation in a population vulnerable to its potential long-term effects.3
Although recent publications have addressed radiation-related topics in CT imaging of specific thoracic diseases,4–10 the approach of thoracic radiologists to the general matter of patient radiation and their strategies for dose reduction are not known. However, such information may help to focus and further enhance already ongoing efforts in this field.9,11,12 The aim of this article is to list and discuss the available solutions to optimise and reduce radiation dose in adult thoracic MDCT examinations.
Methods for Dose Reduction
The term ‘standard dose’ refers to the dose usually recommended by CT manufacturers and often used in routine practice, but that could be substantially reduced – to an optimised dose level – without deleterious effects on image quality. The term ‘optimised dose’ refers to a dose that provides adequate image quality but without excessive radiation, and is the practical application of the ‘as low as reasonably achievable’ (ALARA) principle. The term ‘low dose’ should be restricted to a CT-delivered dose not higher than that delivered by a set of plain films investigating the considered condition. At low doses, image quality is lower but diagnostic accuracy is preserved.
The optimisation process is by definition a process that eliminates the excess of radiation that does not provide a significant increase in image quality. Optimised dose level for a given examination is not known. It has to be defined for a standard human body that may be defined as weighing 70–140kg. This optimised dose level depends on many factors and in particular on the CT technology. The more recent the CT is, the lower the required dose to provide adequate image quality. As there is no large consensus on optimised dose levels, the European regulatory authorities have developed an approach for dose reduction that is based on survey studies.
The unique strategy for reducing the collective radiation dose from diagnostic CT examinations proposed by the EU in 1997 is based on surveys that intend to define a reference diagnostic level (RDL) corresponding to the 75th percentile of the observed doses for a given CT examination (see Table 1). The first historical survey was conducted in the UK on single-detector-row scanners in the mid- 1990s and served as the European RDL across the entire EU. The RDL is expressed either in weighted CT dose index (CTDIw), serving as an index of image quality, or in dose–length product (DLP), expressed in mGy.cm, serving as the indicator of exposure per acquisition and/or per examination in the case of multiphasic procedures. The historical European RDL for chest CT was as high as 650mGy.cm for one single acquisition. The corresponding effective dose (estimating the cancer risk) is calculated by multiplying the DLP by a conversion factor (0.017mSv/mGy.cm) and is 11mSv. The lifetime risk of dying from a radiation-induced cancer if exposed to 11mSv is calculated by multiplying the effective dose expressed in Sievert (Sv) by 5%. For the chest RDL, this lifetime risk is 1/1,820.
According to the EU rules, any radiology department performing CT examinations with a dose higher than the RDL is expected to reduce this dose.15 Thus, after publication of a national survey, the collective dose from CT should decrease because the 25% highest dose values should be reduced. A renewed survey should thus redefine an RDL lower than the original one and the process of dose reduction should be repeated. The first surveys conducted in the EU typically showed that a factor of 2 to 6 was observed between the dose level of a P75 and that of a P25. If a second survey is conducted within a reasonably short time interval after the first one, it is believed that the reference dose levels could decrease significantly and that the interval between DLP corresponding to the P75 and the P25, could be reduced.
Unfortunately, to the best of our knowledge, no EU member has yet performed such repeated surveys and all published national RDL are still very high, ranging from 430mGy.cm in the UK to 627mGy.cm in Italy.13 In 2010 – 12 years after the publication of EU 97/43 – the aims of the directive have fallen a long way short.
Automatic Exposure Control Systems
It is of note that most RDLs were obtained from surveys performed in the late 1990s or the early 2000s. At that time, CT scanners were not equipped with automatic exposure control systems (AECs), also called tube current modulation devices, because these systems were introduced only in 2002.16 AECs are able to provide equalised image quality throughout the helical acquisition for all patients. The technical approach for AEC varies between manufacturers, but all AEC systems are able to automatically adapt the dose to the patient’s size, weight and/or absorption. Thus, AEC can reduce the dose in small patients but also increase the dose in obese patients. Before AECs were introduced, the concern about radiation dose was not as high as it is now. As a matter of fact, standard CT with the same high dose was applied to all patients while providing satisfactory image quality. This means that standard CT delivered a radiation dose suited for obese patients and a significant excess dose in all non-obese patients. One can estimate that the mean amount of excess radiation dose was at least 50% of the delivered one. As AEC systems are now widely used, there are two major problems concerning surveys. First, in order to take into account that AEC systems are widely used, surveys have to be focused on standard patients in order to get rid of the effect of distortion in weight distribution among participants. Second, new surveys suitable for modern MDCT scanners equipped with AECs should be urgently conducted throughout the entire EU in order to definitely abandon the historical RDLs that are approximately twice as high for standard patients and that suited obese patients only.
New Reference Values
Due to this urgent need for new RDLs, an electronic survey has recently been conducted among chest radiologists who are members of thoracic scientific societies around the world.14 This survey revealed that 60% of respondents acquire chest MDCT with DLP <250mGy.cm in a standard patient, more than 80% of them using the AEC device that equips their 8–64 MDCT.14
This DLP value may serve as new reference for thoracic MDCT scanning powered by AECs. Assuming that a chest MDCT covers a longitudinal (caudocranial) distance of 30cm in a standard patient, the reference value (75th percentile) of a chest MDCT expressed in volume CT dose index (CTDIvol) would be <7.8mGy. The median CTDIvol value would be around 6mGy and the 25th percentile, usually presented as the goal for MDCT optimisation, would be around 4.5mGy.13 In addition, as explained below, CT pulmonary angiography can be obtained with a higher vessel enhancement while using a lowered tube potential at 100kV. With 100kV, the CDTIvol is reduced by 30% compared with 120kV. The goal to achieve in terms of CTDIvol in a standard patient undergoing CT pulmonary angiography is thus ≤4mGy. No doubt that the newest scanner generation with technical advances in dose reduction (new filters, new detectors, new reconstruction algorithms such as iterative reconstruction) will enable the reduction of these values by 30 to 70% with equal or even higher image quality (see Figure 1).
Optimisation of a CT radiation dose consists of reducing the dose to the lowest possible level while maintaining image quality at an acceptable and comfortable level. Good competence in CT technique and in particular in the AEC system that is available is required. The main parameter to optimise is the index of image quality. This index is scanner (manufacturer)-specific and usually difficult to manage. Additional parameters to manage may be the minimum and maximum tube current, the slice thickness, the rotation time, the pitch and the reconstruction kernel. Another important parameter to set up is the tube potential. The optimisation process may be facilitated by a simulator of the image that would be obtained with the pre-selected parameters (Toshiba Medical Systems). Optimisation is part of the ALARA concept and could be considered as part of normal daily practice in CT. On the other hand, optimisation that consists of testing on patients could be considered clinical research and may require authorisation by a local ethical committee in addition to written patient informed consent. To date, no guidelines or rules are available in terms of the appropriate methods for dose optimisation and ALARA ‘behaviour’. In addition, as AEC systems are complex and usually not well-known by CT users, the optimisation process is practically very complex.
Many factors contribute to the absence of significant dose optimisation in daily practice. First, as explained above, the regulatory authorities do not provide radiologists with freshly renewed survey data, so RDL are high. Second, the European recommendation is to focus the attention on the third quartile of surveys (P75), but the concept of the 75th percentile is probably insufficient to significantly reduce the collective dose. It has been advised13 to focus on the first quartile (P25). The P25 could be used by CT centres as the goal to be reached, particularly with modern MDCT scanners. Third, there is no penalty for any CT centre that would not reduce an excess in dose. Fourth, AEC are so complex and not sufficiently explained to CT users by manufacturers that radiologists may not feel confident or even may be completely unable to modify the AEC parameters installed by the vendor on their CT. Fifth, radiologists fear to reduce the dose for many reasons, one of these being the absence of training in the use of optimised dose and in the set-up of AEC systems. Sixth, hospital physicists who are familiar with dose measurements are not able to propose clinically relevant noise index levels that fit with all CT scanners, indications and AEC systems. Seventh, the standard set-ups proposed by manufacturers are almost systematically too high by 40–50%, mainly because the manufacturers want to satisfy clients and because these clients (radiologists) usually want excellent image quality without any compromise. Finally, the way radiologists are educated in their universities could have a significant impact on what they are going to promote and use for their entire career.
German, along with French and Belgian, radiologists are more familiar with dose justification, which usually gives a much higher importance to the highest possible detection rate whatever the radiation risks, and to the most perfect image quality. This is at least what the author has personally observed in a recent optimisation process conducted in Luxembourg, a country where radiologists work who have been educated in France, Belgium and Germany.
Low-dose Chest Multidetector-row Computed Tomography
Routine Chest Computed Tomography
The concept of reducing the radiation dose in chest CT was first introduced in 1990 by Naidich et al., who reduced the tube current on incremental 10mm collimation CT and demonstrated that with low tube current settings (i.e. 20mAs), the image quality is sufficient for assessing the lung parenchyma.
While these images are sufficient for assessing lung parenchyma, the increased noise results in marked degradation of the quality of images photographed with mediastinal window settings. Due to this, these authors recommended that such a low-dose technique would be most suitable for children and for screening. As such, these recommendations have been implemented and further studied in lung cancer screening programmes.17
Similar dose-reduction strategies have been applied to thin-section CT, in which no significant difference in lung parenchyma structures was detectable between low dose (i.e. 40mAs) and high dose (i.e. 400mAs).18 Although the observed differences were not statistically significant, changes in ground-glass opacity were difficult to assess at low-dose CT because of the increased noise. Therefore, it was recommended that 200mAs should be used for initial thin-section CT and lower doses (i.e. 40–100mAs) for follow-up examinations.
The relationship between radiation exposure and image quality at mediastinal and pulmonary window settings has been evaluated on conventional 10mm collimation CT images on a single model of CT scanner with mAs settings ranging from 20 to 400mAs.19 Although this study showed a consistent increase in image quality with radiation dose, no difference in detection of mediastinal and lung abnormalities could be detected. These findings were confirmed on MDCT by Dinkel et al.,20 who showed that a 90% reduction in dose compared with standard-dose techniques was not associated with impaired detection of suspicious lesions of malignant lymphoma and extrapulmonary tumours.
In order to investigate the effect of dose reduction without scanning patients several times at several dose levels, it is now possible to use computed simulation of dose reduction by adding random noise to the image obtained at a standard dose. In a validation trial, it has been shown that experienced chest radiologists were unable to distinguish CT images obtained with simulated reduced doses from those obtained with really reduced doses. This technique of simulated reduced doses allows investigators to determine the impact of dose reduction on diagnostic performances without exposing patients to additional radiation and/or several injections of iodinated contrast material.
Computed Tomography Pulmonary Angiography
The simulated low-dose technique has been used to evaluate the effect of dose reduction on CT pulmonary angiography. A group of 21 individuals that showed at least one filling defect within a pulmonary artery was used to simulate CT pulmonary angiography with reduced radiation doses at 60, 40, 20 and 10mAs. This study showed that frequencies of positive and inconclusive results and the branching order of the most distal artery with a filling defect were not changed when tube current–time product was reduced from 90 to 10mAs. On the other hand, the quality of intravascular contrast enhancement decreased when the tube current–time product setting was lower than 40mAs. This study suggests that the reduction of the tube current–time product setting to 40mAs to achieve a reduced radiation dose at CT pulmonary angiography appears to be acceptable.21
Sigal-Cinqualbre et al. have assessed the feasibility of low-kilovoltage in CT pulmonary angiography protocols and have evaluated the effect of such protocols on image quality.22 These authors have simultaneously reduced the tube potential and increased the mAs settings. They have shown that in patients weighing <75kg, 80kV (and 135 or 180mAs, respectively, in patients weighing <60 or 75kg) are sufficient to obtain the same image quality as in patients >75kg and scanned at 120kV and 90mAs. These results need to be confirmed and verified in other indications than CT pulmonary angiography, but this study has already suggested that reducing the tube potential could be a valid method – an alternative to decreasing the mAs settings – to reduce the radiation dose. Since 2004, several studies have validated the used tube potential at 100kV in patients ≤100kg.23 The dose reduction of 100kV acquisition is 30% compared with 120kV. The recommended CTDIvol for standard patients undergoing CT pulmonary angiography is therefore 30% lower than that for a routine chest MDCT and should be as low as 4mGy.
Air Trapping and Expiratory Computed Tomography
By demonstrating air trapping, expiratory thin-section CT is able to detect a disease before the functional tests. This makes this technique an essential part of the diagnosis of bronchiolitis of various origins. As expiratory CT is most often obtained after inspiratory CT, this additional acquisition exposes patients to supplementary radiation dose.
This is of concern in patients with bronchiolitis because they can often be young and, despite their relatively favourable prognosis, they have a high risk of recurrence resulting in repeated follow-up examinations and repeated exposure to CT radiation. In order to investigate the possible effect of dose reduction on the visual quantification of air trapping, we considered the bronchiolitis obliterans syndrome (BOS) after lung transplantation as a model for bronchiolitis.24 In this model, we applied the simulated low-dose technique on expiratory thin-section CT examinations in patients with possible BOS. In 27 lung transplant recipients, expiratory thin-section CT was performed at 140kVp and 80 effective mAs. Dose reduction corresponding to 60, 40 and 20 effective mAs was simulated. This study showed that a simulated dose-equivalent of 25% of the standard dose, i.e. 20mAs, had no substantial effect on the visual quantification of air trapping. As its radiation dose approximated that of incremental thin-section CT with 10mm section intervals performed with a standard dose, expiratory low-dose MDCT could be used in the assessment of air trapping in patients with suspected bronchiolitis. This model could be extended to other origins of bronchiolitis.
CT radiation dose optimisation and reduction is a complex process that seems hardly to have changed over the years. Optimisation behaviour requires strong efforts and close co-operation between radiologists, manufacturers and regulatory authorities for obtaining the significant results that were originally expected from the European Directive 97/43. The chest is an appropriate body region for dose reduction and optimisation because of its natural contrasts. The actual recommended CDTIvol for standard patients undergoing helical chest CT is 6mGy, whereas the corresponding DLP is 180mGy.cm. CT pulmonary angiogram benefits from lowering tube potential from 120 to 100kV and the corresponding typical CTDIvol is at 4mGy, whereby the DLP can be reduced to 120mGy.cm in a standard patient. The newest MDCT generation should enable further dose reduction of 30–70%.