Radiation protection issues in dynamic contrast-enhanced (perfusion) computed tomography.

G Model EURR-6938; No. of Pages 12

ARTICLE IN PRESS European Journal of Radiology xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

European Journal of Radiology journal homepage: www.elsevier.com/locate/ejrad

Radiation protection issues in dynamic contrast-enhanced (perfusion) computed tomography Gunnar Brix a,∗ , Ursula Lechel a , Elke Nekolla a , Jürgen Griebel a , Christoph Becker b a Department of Medical and Occupational Radiation Protection, Federal Office for Radiation Protection, Ingolstädter Landstraße 1, D-85764 Oberschleissheim, Germany b Department of Clinical Radiology, Grosshadern Clinic, Hospital of the Ludwig-Maximilians University, Marchioninistraße 15, D-81377 Munich, Germany

a r t i c l e

i n f o

Article history: Received 19 September 2014 Accepted 11 November 2014 Keywords: Dynamic contrast-enhanced CT Perfusion CT Radiation exposure Health effects Justification Optimization

a b s t r a c t Dynamic contrast-enhanced (DCE) CT studies are increasingly used in both medical care and clinical trials to improve diagnosis and therapy management of the most common life-threatening diseases: stroke, coronary artery disease and cancer. It is thus the aim of this review to briefly summarize the current knowledge on deterministic and stochastic radiation effects relevant for patient protection, to present the essential concepts for determining radiation doses and risks associated with DCE-CT studies as well as representative results, and to discuss relevant aspects to be considered in the process of justification and optimization of these studies. For three default DCE-CT protocols implemented at a latest-generation CT system for cerebral, myocardial and cancer perfusion imaging, absorbed doses were measured by thermoluminescent dosimeters at an anthropomorphic body phantom and compared with thresholds for harmful (deterministic) tissue reactions. To characterize stochastic radiation risks of patients from these studies, life-time attributable cancer risks (LAR) were estimated using sex-, age-, and organ-specific risk models based on the hypothesis of a linear non-threshold dose–response relationship. For the brain, heart and pelvic cancer studies considered, local absorbed doses in the imaging field were about 100–190 mGy (total CTDIvol , 200 mGy), 15–30 mGy ( 16 m Gy) and 80–270 mGy (140 mGy), respectively. According to a recent publication of the International Commission on Radiological Protection (ICRP Publication 118, 2012), harmful tissue reactions of the cerebro- and cardiovascular systems as well as of the lenses of the eye become increasingly important at radiation doses of more than 0.5 Gy. The LARs estimated for the investigated cerebral and myocardial DCE-CT scenarios are less than 0.07% for males and 0.1% for females at an age of exposure of 40 years. For the considered tumor location and protocol, the corresponding LARs are more than 6 times as high. Stochastic radiation risks decrease substantially with age and are markedly higher for females than for males. To balance the diagnostic needs and patient protection, DCE-CT studies have to be strictly justified and carefully optimized in due consideration of the various aspects discussed in some detail in this review. © 2014 Elsevier Ireland Ltd. All rights reserved.

1. Introduction Technical innovations in multi-detector computed tomography (CT) allow for larger volume coverage in shorter scan times and have thus stimulated the application of dynamic contrastenhanced (DCE) CT techniques in clinical practice. Since the temporal change of CT densities (in Hounsfield units, HU) upon

∗ Corresponding author. Tel.: +49 3018 333 2300; fax: +49 3018 333 2305. E-mail addresses: [email protected] (G. Brix), [email protected] (U. Lechel), [email protected] (E. Nekolla), [email protected] (J. Griebel), [email protected] (C. Becker).

administration of an extracellular CT contrast agent (CA) is related to the regional blood supply and the extravasation of the CA, DCECT is a valuable tool for rapid and non-invasive characterization of tissue microcirculation.1 It has been established in medical care for

1 “DCE imaging” is frequently used in the medical literature synonymous with “perfusion imaging”. In this paper, the latter denotes more specifically any imaging technique (CT, MRI, SPECT) that aims primarily at the assessment of tissue blood flow (perfusion) as an important but not the only relevant feature of tissue microcirculation. Accordingly, the term “CT perfusion imaging” specifically refers to a DCE-CT study optimized to acquire serial CT data during the first-pass of a rapidly injected CA bolus through the terminal vascular bed, from which the perfusion can be determined.

http://dx.doi.org/10.1016/j.ejrad.2014.11.011 0720-048X/© 2014 Elsevier Ireland Ltd. All rights reserved.

Please cite this article in press as: Brix G, et al. Radiation protection issues in dynamic contrast-enhanced (perfusion) computed tomography. Eur J Radiol (2014), http://dx.doi.org/10.1016/j.ejrad.2014.11.011


ARTICLE IN PRESS G. Brix et al. / European Journal of Radiology xxx (2014) xxx–xxx

2

improved diagnosis and treatment of cerebral ischemia and infarction [1–4] and is increasingly investigated in clinical trials to define its role in the diagnosis, management, and prognosis of patients with coronary artery disease (CAD, [5–7]) and cancer [8,9]. The most recent state of the art is summarized in this special issue. As compared to DCE magnetic resonance imaging (MRI), the CT technique offers the major methodological advantage of a direct and almost linear relationship between the CA-induced density increase and the local CA concentration. Moreover, it has practical advantages in that the examination is fast, commonly available and better applicable for critically ill and intensive care patients than MRI. But there are also two serious and interconnected shortcomings: First, the relatively small CA-induced increase in CT densities in most (in particular ischemic) tissues as compared to the noise level of the acquired density–time curves and, second, the exposure of patients to ionizing radiation. Although the effective dose resulting from a DCE-CT study is typically less than about 30 mSv, local doses in the examined body region are rather high and may result in harmful radiation damages when the examination is repeated several times or combined with other high-dose angiographic or interventional procedures. Imanishi et al. reported on temporary bandage-shaped hair loss occurring in three patients with cerebrovascular disorders who underwent several CT perfusion studies and two angiographies of the head within a few days [10]. In the United States, approximately 385 patients from six hospitals were exposed to excess radiation during CT brain perfusion scans using inadequate protocols. Some of these patients reported obvious signs of excessive radiation exposure following their scans, such as hair loss or skin redness, which called attention to the problem [11]. But even if radiation exposures are not high enough to produce obvious signs of radiation injury, it can place patients at increased risk for long-term radiation effects, in particular cancer [11]. It is thus the aim of this review article (i) to briefly summarize the current knowledge on deterministic and stochastic radiation effects relevant for radiation protection of patients, (ii) to present the essential concepts for estimating radiation doses and risks associated with DCE-CT studies as well as representative results, and (iii) to discuss in detail the various aspects that have to be considered in the process of justification and optimization of DCE-CT studies as summarized in Fig. 1.

2. Biological effects of ionizating radiation According to how the tissue response relates to the radiation dose, radiological protection deals with two types of adverse health effects. (i) Stochastic radiation effects (cancer or heritable effects due to cell transformation), which may be observed as a statistically detectable increase in the incidences of these effects occurring long after radiation exposure in the affected individuals or their offspring. (ii) Deterministic radiation effects (harmful tissue reactions due to cell killing), which occur at higher doses exceeding tissue-specific thresholds, often of an acute nature [12]. Most recent scientific data question this mainstream classification to a certain extent (see below) and may call for a modified concept for categorizing radiation effects [13,14].

2.1. Stochastic radiation effects Radiation effects at the cellular or molecular level may result in viable but genetically modified cells, which may initiate cancerogenesis in case of somatic cells or may lead to inherited disease in case of germ cells. These all-or-nothing single-cell effects occur by chance (i.e., in a stochastic manner without a threshold), which

implies that their probability but not the severity of the resulting detriment is proportional to the radiation dose. In the context of radiation protection, the main stochastic effect is the occurrence of cancer several years to decades after the exposure has taken place (latency time). Radiation-associated cancers do not differ in their clinical appearance from cancers that are caused by other factors. They can thus not be recognized as such, and it is only by means of epidemiological studies that increases in the spontaneous cancer incidence rates of irradiated groups can be detected. Increased cancer rates have been demonstrated in humans through various radio-epidemiological follow-up studies at organ or whole-body doses exceeding about 50 mGy, delivered acutely or over a prolonged period. The so-called Life Span Study (LSS) of the survivors of the atomic bombings in Hiroshima and Nagasaki is the most important of these studies [15,16]. The follow-up of the atomic bomb survivors has provided detailed knowledge of the relationships between radiation risk and a variety of factors, such as the absorbed dose, the age at exposure, the age at diagnosis and other parameters. The LSS yields cancer incidence [15] and mortality data [16] with good radio-epidemiologic evidence due to the large size of the study population (about 86,600 individuals with individual dose estimates), the broad age- and dose-distribution, the long follow-up period and an internal control group (individuals exposed only at a minute level or not at all). It is, therefore, the major source for predicting radiation-associated risks for the general population. The risk estimates from the LSS are largely supported by a multitude of smaller studies, mostly on groups of persons exposed for medical reasons, both in diagnostics and therapy [17]. Since experimental and radio-epidemiological studies do not provide conclusive evidence for the carcinogenicity of low levels of radiation ( CTDIw ), p > 1 gaps (i.e., CTDIvol < CTDIw ). If the velocity of the patient support, and in turn the pitch, varies during tube loading, the time-averaged CTDIvol is reported. The spiral shuttle mode allows for adapting the scan range to the axial extension of an organ or a tumor (where necessary with an additional margin to take into account respiratory movements). On the other hand, the actual scan length L is larger than the selected range of clinical interest to ensure that the image reconstruction algorithm has sufficient information at the beginning and the end of the body region spirally scanned (“overranging”). The relative contribution of the resulting additional dose to the total dose increases with increasing detector collimation and decreasing axial scan length – and thus poses a particular problem in DCE-CT. For the two spiral-shuttle mode protocols given in Table 1, for example, the actual scan length given by the ratio DLP/CTDIvol is about 30% larger than the defined imaging range. At modern CT systems, the overranging effect can be minimized by dynamic collimation [29]. To characterize the overall absorbed dose delivered to a patient by a specific CT scan, the dose-length product DLP = CTDIvol · L (given in mGy cm) is reported as an additional dose descriptor. In case of serial CT scanning, the CTDIvol and DLP values derived for a single scan have to be multiplied by the total number of frames acquired during the DCE-CT patient study. 3.3. Estimation of organ and effective doses In principle, organ doses can be computed for any CT scan from the CTDIvol and the scan length by Monte Carlo calculations performed for body phantoms approximating the anatomical characteristics of the ICRP reference male and reference female [30]. However, this time-consuming approach is not practicable given the large variety of CT systems and the variability of scan parameters applied in clinical routine. Therefore, organ and effective doses computed for only a few CT systems and parameter settings are used and corrected properly to be applicable to other CT systems and protocols [31]. This task is supported by easy-to-handle software tools [e.g., 32–34]. Alternatively, the effective dose resulting from a CT scan can roughly be estimated from the DLP according to the relation E = kE · DLP by using generic conversion factors kE (in mSv/mGy/cm) reported for several body regions, amongst others for the head (0.0019), chest (0.015), abdomen (0.015) and pelvis (0.013) ([35], for wT factors given in [12]). This approach is very attractive due to its simplicity, but has two shortcomings: (i) The given conversion coefficients were computed for CT scans completely covering the respective body region and can thus, in case of largely different wT factors of the tissues therein, only be used as approximations for CT scans covering a shorter axial range (e.g., the heart region instead of the complete chest). (ii) It does not yield absorbed doses to organs in the primary X-ray beam that are of particular importance for risk assessment of DCE-CT studies. 3.4. Estimation of age-and sex-specific stochastic radiation risks The standard approach to estimate age-, sex-, and organspecific risks resulting from an X-ray procedure is based on radio-epidemiological risk models describing the excess abso¯ T ) (or in a mathematically equivalent lute risk earT (e, a, s, D




5

Table 1 Scan settings of three default protocols implemented at a latest-generation CT system (Somatom Force; Siemens) by the manufacturer for perfusion imaging of the brain, heart, and cancer (trunk). Measurements were performed at an anthropomorphic phantom simulating a slim female (weight, 55 kg; height, 160 cm; BMI, 21.5 kg/m2 ). Parameter settings Shuttle mode Tube voltage (kVp) Tube current modulation (Effective) current–time product per gantry rotation (mAs) Gantry rotation time (s) Detector collimation (mm) Axial length of imaged body region (mm) Thickness of reconstructed slices (mm) Triggering Number of scans Total acquisition time (s) CTDIvol per examination (mGy) DLP per examination (mGy•cm) Effective dosed (mSv)

Brain

Heart

Spiral 70 No 250 0.25 48 × 1.2 114 10 No 32 48 196 2909 6.4

Sequential 80 Yesa 83 0.25 48 × 1.2 100 3 ECG 10b n.s.c 16 172 3.7

Cancer Spiral 100 No 125 0.25 48 × 1.2 114 3 No 32 48 137 2033 32.0

a

Overall adjustment of the tube current to the average body diameter. Typical number of prospectively electrocardiography (ECG)-triggered scans acquired for myocardial perfusion studies in the sequential shuttle mode. c n.s.: not specified. For a given number of scans, the total acquisition time depends on the heart rate. At the scanner considered, ECG-triggered heart scans from two (overlapping) positions of the patient support can be acquired in the sequential shuttle mode every fourth heart beat up to a heart frequency of 110 bpm. d Computed with the tissue-weighting factors given in ICRP Publication 103 [12]. The cancer study was performed in the lesser pelvis. b

manner the excess relative risk) of a person of sex s to develop radiation-associated cancer at age a after exposure at age e to an ¯ T . Estimates of earT (e, a, s, D ¯ T ) for organ-averaged absorbed dose D specific organs are usually based on cancer incidence data of the LSS. In radiation risk models, a linear dose effect relation is commonly assumed for solid tumors, while a linear-quadratic approach usually provides better results for leukemia. The most recent radiation risk models are summarized in the BEIR-VII report [36]. Details of the BEIR-VII-based risk models used in this paper to estimate site-specific radiation-associated cancer incidences can be found in [37]. The organ-specific excess absolute lifetime risk or lifetime attributable risk, LAR, is than calculated by summing up all ¯ T ) values between e + t (with t being the minimum earT (e, a, s, D latency period) and the age of, e.g., 85 years, commonly used for lifetime risk estimates. However, competing risks must be taken into account in this calculation. Describing the conditional probability that a person of age e survives beyond the age a by P(e,a), the organ-specific LAR is given by

85

¯ T) = LART (e, s, D

¯ T ) · P(e, a) da. earT (e, a, s, D a=e+t

The minimum latency period t is the minimum time during which radiation-associated cancer typically does not show clinical symptoms. A t of about 5 years for solid cancer and about 2 years for leukemia is widely applied for incidence data. To determine the total LAR, all organ-specific LART estimates have to be summed up. 3.5. Radiation doses and risks to patients from representative DCE-CT protocols To demonstrate the application of the presented concepts in DCE-CT and to give typical dose and risk values, three representative protocols are considered in detail. They were optimized by the manufacturer for perfusion imaging of the brain, heart and cancer (in the present study of a pelvic tumor [38–40]) at a latestgeneration CT system (Somatom Force; Siemens Healthcare Sector, Forchheim, Germany) in the sequential or spiral shuttle mode. The scan settings are given in Table 1. Dose measurements were performed for these DCE-CT protocols at an anthropomorphic phantom (ATOM dosimetry phantom 702-D; CIRS, Norfolk, Virginia, USA) simulating a slim female

(weight, 55 kg; height, 160 cm; BMI, 21.5 kg/m2 ). The phantom is constructed of proprietary tissue equivalent materials and transected into transaxial cross sections (thickness, 2.5 cm) with holes for thermoluminescent dosimeters (TLDs). Absorbed doses were measured inside and at the surface of the phantom with lithium fluoride (TLD-100; Bicron-Harshaw, Cleveland, Ohio, USA) rods (size, 1 mm × 1 mm × 6 mm) and chips (size, 3.2 mm × 3.2 mm × 0.9 mm), respectively. Individual calibration, annealing, and read-out of the TLDs are described in detail in [31]. Fig. 2a–c shows the transaxial distribution of absorbed doses in representative phantom cross sections. The dose distributions measured in the head and pelvis demonstrate the well-known decrease of dose levels from the periphery to the center. This is particularly marked in case of the head, where the dose plot shows a steep dose gradient at the periphery caused by the strong absorption of X-rays by the skull. (Since this effect is not simulated by the standard head CT dosimetry phantom, the CTDIvol overestimates the real exposure to the brain.) The observed high radiation exposure to the lens of the eye needs particular attention to minimize the risk for radiogenic cataracts [41]. In contrast, the dose distribution obtained for the myocardial perfusion scan does not show markedly higher dose values in the periphery as compared to the center due to the use of a bow-tie beam shaping filter optimized for heart examinations, which limits amongst others the dose to the female breasts. This requires, however, to properly center the patient on the patient support. The three investigated exposure scenarios exemplify that the total (effective) CTDIvol given at the console of the CT system roughly indicates the absorbed doses although there are considerable differences in detail due to the mentioned specifics of the scans. With regard to the prevention of harmful tissue reactions, the following important aspects have to be considered in the assessment of the dose levels given in Fig. 2a–c: Absorbed doses for the investigated heart protocol were measured with tube-current modulation at a slim female and thus considerably underestimate doses in more heavyset patients undergoing the same examinations. Moreover, the dose values characterize radiation exposure in a phantom section outside the border-area between the two alternate sub-scans acquired in the sequential shuttle mode. In body regions where the sub-scans overlap, absorbed doses are increased to some extent. Fig. 2a indicates that only four cerebral perfusion studies performed with the examined DCE-CT protocol will yield absorbed doses to the brain and the lens of the eye exceeding 0.5 Gy


G Model EURR-6938; No. of Pages 12 6


absorbed doses in the imaged body region (head: 100–190 mGy; heart: 15–30 mGy, pelvis 80–270) due to the strongly differing wT factors of the exposed tissues. This substantiates that the effective dose is per definition not at all predictive of local absorbed doses resulting from DCE-CT and that a limit on this quantity may not necessarily protect against radiation-induced local deterministic tissue reactions. Using the presented approach for risk estimation as well as German disease and life table data [42,43], total LARs were estimated (with a dose and dose-rate effectiveness factor of 1) for the considered DCE-CT examinations. The estimates are plotted in Fig. 3 for both sexes and different ages at exposure. As a general result, the plots demonstrate that the LAR to develop a radiation-associated cancer decreases markedly for all examinations with increasing age at exposure and is – in particular at younger ages – markedly higher for females than for males. Due to the high radiation susceptibility of the female breast and lung, radiation risks estimated for the examined cerebral and myocardial perfusion studies are comparable although the absorbed doses are substantially higher in the head as compared to the heart region. The LARs estimated for the two exposure scenarios are less than 0.07% for males and 0.1% for females at an age of exposure of 40 years. This corresponds to 1 excess cancer in approximately 1400/1000 examined males/females. For the considered tumor location and protocol, the corresponding risks are more than 6 times as high. Recalling the continuing uncertainties in modeling radiationassociated cancer risks, the given LAR values should be interpreted with a necessary sense of proportion as conservative numbers in accordance with the precautionary principle. In the patient briefing, they can be compared with the lifetime baseline cancer risk, i.e. the “normal” risk to develop cancer during lifetime, and put into perspective herewith. In Europe, the lifetime baseline cancer risk for all cancers (excluding skin cancer) is about 50% for men and 40% for women (estimated from the age- and sex-specific incidence cancer rates given in [44] for a multitude of European countries). 4. Justification of DCE-CT studies Current radiation protection recommendations and directives [e.g., 45–47] require that

Fig. 2. Transaxial distributions of absorbed doses (in mGy) measured by TLDs at an anthropomorphic phantom simulating a slim female placed in supine position (with the head not tilted toward the chest as suggested for patient examinations) on the patient support of a latest-generation CT system (Somatom Force). Perfusion imaging of (a) the brain (70 kVp, total CTDIvol = 196 mGy), (b) the heart (80 kVp, total CTDIvol = 16 mGy) and (c) a pelvic tumor (100 kVp, total CTDIvol = 137 mGy) was performed using the default DCE-CT protocols specified in Table 1 in more detail. The numbers in italic type given in (b) were measured in the breasts at the axial position of the shown transaxial phantom section. Whereas the dose distributions are approximately bilaterally symmetric, doses are considerably lower at the posterior as compared to the anterior side of the anthropomorphic phantom because of the absorption of X-rays in the patient support when the tube is below the table. For other values of the current–time product per scan and/or number of serial CT scans (but otherwise identical parameter settings), the given dose values have to be scaled by the ratio of the respective total current–time products.

above which cerebrovascular disorders and opacification of the lens (when they are in the primary X-ray beam) become increasingly important. Effective doses to patients estimated for the three DCE-CT examinations are about 6.4 mSv (head), 3.7 mSv (heart) and 32 mSv (pelvis) and do not correctly indicate the markedly different

- all new types of practice with a specified clinical objective involving exposure of patients to ionizing radiation (e.g., DCE-CT for the diagnosis of cerebral ischemia) shall be justified in advance before being generally adopted (generic justification); if a type of practice is not justified in general, it can be justified in special circumstances based on a documented case-by-case decision or in clinical trials reviewed by an ethics committee; - all individual medical exposures shall be justified in advance taking into account the specific objectives of the exposure and the characteristics of the individual involved as well as the efficacy, benefits and risks of available alternative techniques having the same objective but involving no or less exposure to ionizing radiation. 4.1. Generic justification The generic justification of a defined type of practice is a matter for national and international professional bodies, in conjunction with national health and radiological protection authorities, based on the results of clinical trials [45]. In this context, the severity of the disease to be diagnosed and treated as well as the prognosis of the patients have to be taken into account. According to the “Heart Disease and Stroke Statistics” reported by the American Heart Association [48], within 5




7

Fig. 3. Lifetime attributable risks (LAR) for the DCE-CT protocols specified in Table 1 for perfusion imaging of the brain, heart, and a pelvic tumor at different ages in adult (a) male and (b) female patients. Risks were estimated using the most recent BEIR-VII models as well as German life tables and cancer incidence rates. For other values of the current–time product per scan and/or number of serial CT scans (but otherwise identical parameter settings), the given risk values have to be scaled by the ratio of the respective total current–time products.

years after a first myocardial infarction at 45–64 (≥65) years of age, 5% (25%) of men and 8% (30%) of women will die. The proportion of patients dying within 5 years after a first stroke at 45–64 (≥65) years of age is 26% (55%) for men and 29% (57%) for women. These relative 5-year mortality rates for (white) patients suffering from myocardial infarction or stroke are in the middle range of data reported for various cancer sites that range in Europe from less than 10% for testicular cancer to nearly 90% for pancreas cancer [49]. Overall, CAD (20/22% in men/women), stroke (10/15%) and cancer (23/19%) are the leading single causes of death in Europe [50]. Therefore, any substantial progress in the diagnosis, management and prognosis of these diseases, e.g. by justified DCE-CT protocols, will have a significant net benefit not only for the patients affected (e.g., a reduced long-term disability and increased survival) but also for the society (e.g., decreased overall health care and economic costs) – particularly given the fact that imaging-related cancer risks are usually relatively low (although not completely negligible) as compared to the patients’ baseline cancer risk and the lethality risk from an inadequate treatment. It should be noted that the LARs plotted in Fig. 3 for three representative DCE-CT protocols are estimated by using life table data for the general population and not by using data specific for patients suffering from circulatory diseases or cancer and thus overestimate the stochastic radiation risk due to the reduced life expectancy of these patients. For the large proportion of patients having a survival time shorter than the latency period for the development of a clinically manifest radiation-associated cancer, stochastic radiation effects can even be ignored completely. There is a growing number of studies demonstrating a benefit of DCE-CT for the diagnosis and management of patients suffering from the mentioned severe diseases. Based on the available clinical evidence, cerebral perfusion CT has already been established in medical care [1–3] and guidelines for the performance of the examinations have been issued [51,52]. In contrast, the specific role and indications of DCE-CT in clinical cardiology and oncology has not been determined definitely [5–9,53]. Although preliminary studies yielded promising results, several methodological problems need to be resolved and larger patient studies need to be performed before DCE-CT can be implemented in routine practice to noninvasively assess the vascular support of the myocardium or cancer. Nevertheless, there may be clinical circumstances that justify on a case-by-case decision the application of DCE-CT in patients with CAD or cancer although the procedure is not yet justified in general.

4.2. Individual justification In the clinical setting, each single DCE-CT study of a patient should be carefully justified to indicate that the clinical benefits outweigh the risks and alternative investigations have been considered that do not use ionizing radiation. An exception is the use of DCE-CT in medical research, where the individual justification is replaced by an examination of the study by an ethics committee and/or the competent authority. The following discussion addresses four key questions regarding the individual justification of a DCE-CT study and presents important (but by no means complete) associated considerations to exemplify the patient-specific decision-making process. Is the intended DCE-CT examination likely to provide new information on tissue microcirculation that will affect diagnosis and therapy management of the individual patient, and is the radiation risk of the procedure less than the risk of missing that information? Answering these questions requires an adequate communication and consultation between the radiologist and the referrer possessing the relevant clinical information. In most clinical situations, the pre-test probability of the disease being tested for is an important factor that determines the clinical relevance of the results of an intended DCE-CT examination. For example, myocardial perfusion imaging is best performed in patients with intermediate pre-test probability of developing obstructive CAD; it is in general not indicated in low-risk patients [54]. As a matter of course, a DCE-CT examination is by no means justified when the necessary information was already gained during the clinical work-up of the patient in another facility or with another imaging modality. What microcirculatory tissue parameters are required for diagnosis, prognosis, therapy planning and/or monitoring? This is an important aspect since it determines the total number of CT frames to be acquired (see below) and thus strongly affects patient exposure to ionizing radiation. In patients with acute ischemic stroke, for example, it is standard practice to map the tissue perfusion (regional cerebral blood flow) to assess the central core of ischemia with a substantially reduced flow that will lead to cell death when circulation is not re-established in time. When brain tissue potentially at risk of infarction should be visualized in the peri-infarct region or in patients suffering from transient ischemic attacks (TIA) or prolonged reversible ischemic




neurologic deficits (PRIND), the regional cerebral blood volume is also of relevance, since the reduced blood pressure in brain territories affected by a vessel occlusion is autoregulated as far as possible via a vasodilation of microvessels to reduce the vascular resistance and thereby to maintain the blood flow [55]. Even more conclusive, the cerebrovascular reserve capacity of tissues at risk can be evaluated by a pharmacologically induced stress test using a vasodilator. The same considerations apply to myocardial perfusion studies. For the characterization of tumor microcirculation and angiogenesis, which are becoming increasingly important for individualized treatment planning and monitoring, not only the blood flow and volume are of relevance but also the (substance-specific) leakiness of tumor microvessels [9,56]. The pharmacokinetic properties of conventional CAs used for DCE-CT studies make them appropriate probes to assess spatially resolved the extravasation of circulating low-molecular-weight anticancer drugs into the interstitial space of tumors [57]. Are alternative imaging tests involving no or less exposure to ionizing radiation available and appropriate to get the required clinical information? In the assessment of patients with acute stroke, DCE-CT and DCE-MRI provide (usually in combination with a CT or MRI angiography, respectively) similar information [4]. As compared to MRI, however, CT has logistic advantages in the initial evaluation of patients with suspected acute stroke since it is widely available round-the-clock in most medical institutions and fewer patients have contraindications to it. This is of particular relevance when evaluating stroke patients in the emergency setting who are in the therapeutic window for thrombolytic therapy [51]. In this situation, CT very quickly allows the essential differentiation between ischemic (about 85%) and hemorrhagic stroke (about 15%) [51,58]. For therapy monitoring as well as for the assessment of patients with TIAs or PRINDs, however, MRI (angiography as well as perfusion and diffusion imaging) should be preferred if at all possible. Coronary CT angiography has a high reliability in the exclusion of CAD (i.e. a high negative predictive value), but severity of stenosis correlates poorly with ischemia so that anatomy-based decision-making with respect to revascularization is limited [54]. Therefore, CT angiography has to be complemented by myocardial perfusion imaging to demonstrate not only an obstructive coronary artery stenosis but also the location and extent of a potential myocardial ischemia. At present, myocardial perfusion imaging is frequently performed in clinical practice by single-photon emission computed tomography (SPECT), which allows a robust rest and stress examination performed on one day or two. Using Tc-99mlabeled sestamibi or tetrofosmin as radiopharmaceuticals, the total effective dose for a combined rest and exercise study is between 4 and 6 mSv [59] and thus slightly lower than the total effective dose of a rest and stress CT perfusion scan [7] of the whole myocardium (cf. Table 1). The differences in the absorbed doses to the breast, lung and heart are even more pronounced in favor of SPECT. On the other hand, a low-dose CT scan is additionally acquired when patients are examined at hybrid SPECT/CT systems to allow for a reliable attenuation correction and, in turn, the quantification of myocardial tracer uptake. To reduce radiation exposure of stable CAD patients, the stress study can be performed first, so that the rest study requiring a separate tracer injection may be omitted if the stress images are entirely normal. In contrast, in patients with acute coronary ischemia syndromes only the rest study is performed. Quantitative analysis of attenuation-corrected SPECT and DCE-CT data may enable the use of stress-only protocols for the diagnosis and prognosis of CAD [60]. A limitation of cardiac SPECT is the very long time (several hours) needed for the entire procedure (fasting before tracer injection, wait between tracer injection and data acquisition as well as between the exercise and rest study if indicated) [61]. For reasons of the patient’s critical health condition

and/or compliance or timeliness of results, a DCE-CT examination combining an anatomical and functional imaging test (“one-stop shop”) may be justified as an alternative in specific situations, while fully respecting the still missing generic justification of DCE-CT for diagnosis and management of CAD patients. In any case, however, myocardial perfusion imaging (and coronary angiography of proximal coronary arteries) with MRI [54] has to be preferred whenever possible since it does not expose patients to ionizing radiation. Despite the fact that CT has been established as the main clinical tool for tumor staging and therapy monitoring, the general considerations presented at the end of the previous paragraph also hold for the functional characterization of the angiogenesis state of a tumor by DCE-CT – in particular in case of young and middle-aged patients suffering from solid cancer with a favorable prognosis (e.g. prostate, uterus, or breast cancer). Due to the reasons discussed above, a higher radiation dose may in contrast be acceptable for patients of advanced age or a cancer with a poor prognosis (e.g. pancreas or lung cancer). Was the patient already exposed to ionizing radiation for diagnostic or therapeutic purposes or are such exposures planned? When a DCE-CT examination of the same body region is repeated several times for therapy monitoring or combined with other high-dose angiographic or (in particular neuro-) interventional procedures, harmful tissue reactions such as radiogenic cataracts, cerebro- and cardiovascular diseases or, in extreme cases, skin reactions may occur. To avoid deterministic and to limit stochastic radiation effects, each single DCE-CT examination should be justified as far as possible in the context of all other procedures already performed or planned. Again, this requires an adequate communication between the various medical experts involved in the diagnosis and treatment of the individual patient. For patients scheduled for or already undergoing a radiation therapy, the radiation dose resulting from one or even a few DCE-CT examinations is low as compared to the dose in the therapeutic irradiation field(s). Nevertheless, the radiation therapist should be informed of the additional imaging-related exposure, at least when several DCECT examinations are scheduled before, during or after the course of radiation therapy.

5. Optimization of DCE-CT studies Once a specific imaging procedure has been justified for an individual patient, it has to be optimized, which means that patient exposure is kept as low as reasonably achievable (ALARA principle) consistent with obtaining the required medical information, taking into account economic and societal factors [45–47]. The primary objective of optimizing a DCE-CT study is to set up the examination protocol in such a way that the microcirculatory tissue parameter(s) justified in advance can be estimated (and mapped) with sufficiently high accuracy (low bias or systematic error) and precision (low variance or statistical error) by tracerkinetic analysis from a minimum set of serial CT scans acquired at the least possible dose. When the accuracy is low, estimated tissue parameters are not comparable with data obtained by other institutions using different DCE-CT protocols or imaging modalities such as DCE-MRI. In case of a low precision, parameter estimates vary randomly to a considerable degree from measurement to measurement and may thus not even be appropriate for therapy monitoring when using the same protocol. Basically, four different concepts are used for tracerkinetic analysis of density–time curves derived from DCE-CT patient studies (Fig. 4): the steepest-gradient method [62,63], the indicator dilution theory based on a deconvolution approach [64,65], a closed-form adiabatic approximation to the tissue heterogeneity model [66,67], and two-compartment exchange models [57,68,69].


G Model

ARTICLE IN PRESS

EURR-6938; No. of Pages 12

G. Brix et al. / European Journal of Radiology xxx (2014) xxx–xxx

300

200 150 100

start of i.v. CA injection

tissue-feeding artery 250

9

here – although it can have implications for the generic justification of a particular DCE-CT technique. The following considerations exclusively relate to the a-posteriori (practical) problem of parameter identifiability from noisy and scarce DCE data that needs to be addressed by an optimization of the entire measurement process (image contrast and noise, timing) to get accurate and precise parameter estimates by tracerkinetic modeling of measured DCECT data, and thus has a direct impact on patient exposure to ionizing radiation. 5.1. Scan parameters and image reconstruction

0 14 12 10 8

injection duration lag time

CT [HU]

50

6

steepest gradient 4 2

tissue of interest

0

0,00 0,25 0,50 0,75 1,00 1,25 1,50 1,75 2,00 2,25 2,50

t [min] Fig. 4. Schematic diagram of the timing of a DCE-CT study. For tracerkinetic analysis of tissue microcirculation, density–time courses have to be acquired before, during and after i.v. injection of a CT contrast agent (CA) from both a tissue-feeding artery (top) and the tissue region of interest (bottom). If only the perfusion (blood flow) has to be estimated, as for example by the steepest-gradient method, it is sufficient to acquire serial CT data with a high sampling rate over a short period of time (in the plot 32 scans every 1.5 s). In order to characterize tissue microcirculation in more detail, e.g. by the capillary permeability and the relative volumes of the intravascular and interstitial distribution spaces of the administered CA in a tumor, additional CT scans must be acquired, although with a reduced temporal sampling (in the plot 8 scans every 12 s).

The first approach makes it possible to identify the tissue perfusion (blood flow) from a minimum number of serial CT scans (typically between 30 and 40). The most detailed information on tissue microcirculation is obtained by the two latter techniques which make it possible to derive not only the blood flow and volume but also the capillary permeability and the volume of the interstitial distribution space. However, the price to be paid for this more comprehensive characterization is that serial CT scanning has to be performed over a longer period of time (Fig. 4), which in turn increases the total number of scans that have to be acquired, and thus patient exposure. A detailed description of the various tracerkinetic concepts as well as of their specific advantages and limitations can be found in [56,70–72]. The accuracy and precision with which physiological tissue parameters can be estimated from DCE-CT data by tracerkinetic modeling is closely linked with the more general issue of the a-priori and a-posteriori identifiability of tissue microcirculation by specific measurement techniques. The problem of a-priori (structural) identifiability deals with the uniqueness and formal correctness of tissue parameters determined by tracerkinetic modeling from ideal data. With regard to DCE-CT data this issue has been investigated in some detail in [70] and will not be further addressed

The contrast-to-noise ratio of density–time curves derived from a DCE-CT patient study (Fig. 4) is – apart from the temporal sampling – the most important quantity determining the precision of physiological tissue parameters estimated by tracerkinetic analysis of the data. In particular the tissue perfusion, which is determined from only a few CT values measured during the very early phase of CA uptake, is very sensitive to noise [69]. At a sufficiently high temporal sampling (see below), the maximum CA-induced contrast enhancement in a tissue of interest should be, as a rough rule of thumb, at least tenfold higher than the noise level to allow a reasonable estimation of this tissue parameter [63]. The contrast-to-noise ratio of DCE-CT data and the radiation dose resulting from their acquisition are affected by the following scan and image reconstruction settings that have to be selected with due care. Tube voltage. In CT, contrast, noise and radiation dose depend in a complex manner on the setting of tube voltage. Systematic investigations of the contrast-to-noise ratio obtainable per unit of dose were reported by Kalender et al. for a wide range of photon energies and patient cross-sections [73]. For iodine imaging, the optimum was found at voltages far below 80 kV, which is not practicable at present due to technical limitations. Physically, a reduction of the tube voltage shifts the mean photon energy closer to the k-edge of iodine. At present CT systems, a tube voltage of 100 kV is recommended for DCE-CT of the trunk in most patients (BMI above 25 kg/m2 ) and 70–80 kV for scans of the trunk of slim patients and the head. As compared to the standard voltage of 120 kV, these settings not only result in a markedly improved contrast-noise ratio but also substantially reduce patient exposure [7,9,73,74]. Tube output. All CT manufacturers have developed effective tools for tube-current modulation, to adjust the tube output to both the size and shape of the individual patient [74,75]. This is achieved (i) by an overall adjustment of the current by comparing the size of the actual patient on the basis of a measured topogram with a “standard-sized” patient, and (ii) by additional variation of the current during the rotation of the CT measuring system by realtime monitoring the attenuation of the X-ray beam in the patient at different projection angles during the scan. However, when imaging more heavyset patients, automatic tube-current modulation can result in an effective current considerably exceeding the reference value set for a “standard-sized” patient. In case of DCE-CT studies, this up-regulation mechanism may result in unacceptable high patient exposures. It may thus be more appropriate to perform DCE-CT studies with a tube current fixed to a relatively low value, by deliberately accepting a higher noise level for heavyset patients that can be reduced after data acquisition by means of the techniques described below. This approach can be combined with an angle-dependent tube current modulation to reduce the dose to a radiosensitive organ, such as the female breast or the lens of the eye by reducing the X-ray tube output in real time when the tube is directly in front of these organs. As has been shown, organ-based tube-current modulation yields a dose reduction to the female breast similar to that achieved with bismuth shielding (see below) without adversely affecting image noise or CT density accuracy [76].




It is important to note that the dose is linearly related with the effective current–time product, whereas the quantum noise is inversely related with the square root of this quantity. This means, for example, that doubling the current–time product increases the dose by a factor of 2 but reduces the quantum noise only by a √ factor of 1/ 2 = 0.71. Due to this adverse relationship it is often not possible to find an acceptable balance in DCE-CT between the opposing requirements of limiting patient exposure and image noise by solely adjusting the current–time product. A solution to this dilemma is to use more dose-efficient image reconstruction algorithms and/or to substantially reduce the spatial resolution of reconstructed CT images (see below). An alternative increase of the CA dosage for DCE-CT is generally not feasible without substantially increasing the risk of CA-associated adverse reactions. Pitch. When performing a DCE-CT study in the spiral shuttle mode, patient exposure can be reduced at the expense of spatial resolution – just as in conventional diagnostic CT scanning – by choosing a pitch factor greater than 1 where it is clinically appropriate. Scan length. For DCE-CT studies of patients with cerebral or myocardial ischemia/infarction, the complete organ is generally scanned in the spiral or sequential shuttle mode, respectively. In contrast, the axial body region to be imaged in cancer patients can and should be adapted to the position and extension of the tumor by using the spiral shuttle mode for data acquisition, referring to any previously acquired images, in particular in follow-up studies. Image reconstruction. Iterative image reconstruction has become a hot topic in the past years and is aimed primarily at noise reduction [74,77,78]. Despite promising clinical results, however, the application of this innovative tool is still limited in clinical practice – partly because of an unaccustomed image impression as compared to standard filtered back reconstruction and partly because of the high costs for the required soft- and hardware. Further improvement of the contrast-to-noise ratio of serial images may be obtained without sacrificing the spatial resolution by using highly constrained reconstruction algorithms that take advantage of prior image data (e.g., PICCS, [79]). However, when applying non-standard algorithms for reconstruction of DCE-CT images, it is vital to determine whether the algorithm chosen and the respective reconstruction parameters yield accurate CT densities. Spatial resolution. Regardless of what reconstruction method is used, a reduction of image noise can be achieved either by decreasing the matrix size and increasing the slice thickness of the reconstructed images or by image filtration. Adapting the spatial resolution (not to be confused with the voxel size) of DCE-CT images by these strategies to that of competing imaging techniques yielding comparable physiologic tissue information (MRI, SPECT, PET), will improve the precision of the physiological tissue parameters estimated for each CT voxel to a level that allows at least visual inspection of the resulting parameter maps to identify tissue regions with abnormal microcirculation [80]. Based on the computed parameter maps it is then, in a second step, possible to average voxel values over functionally homogeneous tissue regions to get parameter values with a sufficiently high precision for reliable quantitative characterization of tissue microcirculation.

5.2. Timing Both the accuracy and the precision of physiologic tissue parameters to be estimated from a DCE-CT patient study crucially depend on the timing of the measurement procedure. Therefore, the temporal sampling and total number of serial CT scans as well as the timing of CA administration have to be carefully optimized – bearing in mind that patient exposure is directly proportional to the total number of acquired CT scans.

Temporal sampling. As already mentioned, the most critical parameter is the tissue perfusion that is determined from the initial (or perfusion) phase of contrast enhancement during the first-pass of the CA. Using the frequently applied steepest-gradient technique, for example, the perfusion is approximately given by the maximum up-gradient of the tissue curve divided by the maximum of the arterial input function (cf. Fig. 4). Regardless of what specific tracerkinetic method is used for computing the tissue perfusion, the initial CA uptake must be sampled with a sufficiently high temporal rate to accurately characterize the shape of density–time curves acquired from a tissue-feeding artery and the tissue of interest. It is recommended to acquire DCE-CT data during the perfusion phase with a sampling rate of no less than 1 image every 2 seconds, if procurable 1 image every second [9,52]. In most clinical situations, this requirement is readily achievable even in the shuttle mode at modern CT systems. An exception are prospectively electrocardiogram (ECG)-triggered studies of the heart performed in the sequential shuttle mode, where data required for the reconstruction of CT images covering both body positions have to be acquired in most patients over four cardiac cycles. The resulting temporal sampling has to be considered as borderline with respect to an accurate quantification of myocardial perfusion, even though the arterial input function can be sampled from the aorta from each sub-scan and thus with a rate twice as high compared to the myocardial region of interest. If it is clinically justified to characterize tissue microcirculation in more detail, e.g. by also determining the capillary permeability and the relative volumes of the intravascular and interstitial distribution spaces of the administered CA, additional DCE-CT data have to be acquired during the interstitial (or extravasation) phase. This can and should be done with a reduced temporal sampling of 1 image every 10–15 s (cf. Fig. 4) due to the markedly slower temporal change of contrast enhancement during this distribution phase. Total acquisition time. The period of time over which serial CT scanning has to be performed after rapid CA administration should be as short as possible, taking into account the following (predominantly patient-specific) factors (cf. Fig. 4): (i) The time delay between the start of CA injection and its arrival in the major artery from which the arterial input function is derived. This delay varies with the side of peripheral CA injection and the patient’s cardiac output. (Note: DCE scanning has to be started prior to CA arrival in the reference artery in order to acquire at least one baseline scan required for computing the contrast enhancement in blood). (ii) The duration of CA injection that directly affects the time the arterial input function rises from zero to its maximum. (iii) The lagtime between CA arrival in the reference artery and in the tissue of interest that may be increased substantially in patients with a significant stenosis of a downstream artery supporting the tissue being examined. (iv) A reduced perfusion in ischemic tissues that results in a slow contrast enhancement in the perfusion phase. For these reasons, the total acquisition time should span a total of 30–50 s for first-pass perfusion studies, where the (borderline) duration of 30 s is used for DCE-CT examinations of the heart that have to be performed during a single breath hold. Even shorter acquisition times substantially increase the risk of truncated und thus not evaluable DCE-CT data, which would mean that the procedurerelated radiation exposure to the patient was completely useless. For a more comprehensive characterization of tissue microcirculation (see above), additional DCE-CT scans have to be acquired in the interstitial phase over another 60–120 s (cf. Fig. 4), depending upon the specific model used for tracerkinetic analysis. 5.3. Reduction of patient and organ movement Quantitative analysis of DCE-CT studies is seriously compromised by patient and organ movements and, at the worst, fails




completely. It is thus essential to reduce motions as far as possible to ensure that patients benefit from the study and are not unnecessarily exposed to ionizing radiation. To reduce the adverse effect of patient movements, suitable positioning aids should be used to fixate the body region (e.g., the head) to be imaged. Respiratory motion poses a particular problem for DCE-CT studies. Serial CT scanning of the chest and heart is best performed during breath-holding (scanning of perfusion phase) or shallow breathing (scanning of perfusion and interstitial phase). For DCE-CT studies of the head, neck, lower abdomen and pelvis, it is normally sufficient to instruct the patient to breathe quietly [9]. Prospective ECG-triggering is mandatory for myocardial perfusion studies, even though it results in a suboptimal temporal sampling (see above). In all cases, remaining distortions can be compensated for by appropriate postprocessing tools for 3D matching of serial CT images.

11

calculated for the considered studies are – even if they are repeated a few times – small as compared to the patient’s baseline cancer risk and to the lethality risk from an inadequate treatment of the mentioned severe diseases. In any case, each individual DCE-CT study has to be strictly justified in advance taking into account the various aspects discussed in some detail in this paper. Just as important are measures to carefully optimize the applied DCE-CT protocols in order to ensure that the clinically required physiological tissue parameters can be estimated with necessary accuracy and precision from time-resolved CT data acquired at the lowest achievable radiation dose consistent with this demand. Acknowledgements We thank C. Klingele, E. Klotz, and A. Schegerer for helpful discussions.

6. Protective shielding References DCE-CT examinations can result in substantial radiation doses to radiosensitive tissues near the body surface, such as the lenses of the eye, thyroid, female breast, or testicles. These tissues can either be exposed by low-energy scattered photons due to their proximity to the imaged body region (e.g., the lenses of the eye and the thyroid in case of brain scans) or by primary and scattered photons when they lie in the imaged body region (e.g. the female breast in case of heart scans). A scanner inherent shielding of tissues near the body surface can be achieved by examination-specific bow-tie beam shaping filters. Out-of-plane shielding. Radiation scattered by the CT measuring system or the covering gantry and leakage radiation from the X-ray tube can be reduced by protective shielding of sensitive tissues outside the imaging field. Although the major contribution to the out-of-plane dose comes from internally scattered photons, it is (following the ALARA principle) advisable to protect sensitive tissues near the scanned body region by bismuth shields against external radiation [81]. In-plane shielding. For DCE-CT studies, selective protection of radiosensitive tissues near to the body surface inside the imaging field by bismuth shields is discouraged, because in-plane shields can adversely affect image noise and CT number accuracy [82,83]. As has been shown, organ-based tube current modulation (see above) yields a dose reduction to the female breast similar to that achieved with bismuth shielding, yet without negative effects on image quality [76]. Protection of the eye lenses is best performed by tilting the patient’s head toward the chest whenever possible to reduce direct exposure. 7. Conclusion DCE-CT studies are increasingly used in medical care and clinical trials to improve diagnosis and therapy management of the most common life-threatening diseases: stroke, CAD and cancer. As exemplified by representative DCE-CT protocols, local absorbed doses in the scanned body regions can be rather high and can result in harmful radiation damages (note: the threshold for the cerebro- and cardiovascular systems may be as low as 0.5 Gy) when the examination is repeated several times or is combined with other high-dose diagnostic or therapeutic procedures within a short period of time. Local absorbed doses caused by a DCE-CT study are roughly indicated by the CTDIvol , whereas the effective dose is a generic measure of stochastic radiation risks. For an individual patient, the stochastic radiation risk to potentially develop cancer years after a DCE-CT study has to be estimated by radioepidemiological risk models using organ doses and taking into account both the patient’s age and sex. Lifetime attributable risks

[1] Hochberg AR, Young GS. Cerebral perfusion imaging. Semin Neurol 2012;32:454–65. [2] Biesbroek JM, Niesten JM, Dankbaar JW, Biessels GJ, Velthuis BK, Reitsma JB, et al. Diagnostic accuracy of CT perfusion imaging for detecting acute ischemic stroke: a systematic review and meta-analysis. Cerebrovasc Dis 2013;35:493–501. [3] Dubey P, Pandey S, Moonis G. Acute stroke imaging: recent updates. Stroke Res Treat 2013;76721:2, http://dx.doi.org/10.1155/2013/767212. [4] Sillanpää N. Role of perfusion imaging in acute stroke. Panminerva Med 2013;55:29–41. [5] Becker A, Becker C. CT imaging of myocardial perfusion: possibilities and perspectives. J Nucl Cardiol 2013;20:289–96. [6] Loewe C, Stadler A. Computed tomography assessment of hemodynamic significance of coronary artery disease: CT perfusion, contrast gradients by coronary CTA, and fractional flow reserve. J Thorac Imaging 2014;29:163–72. [7] Rossi A, Merkus D, Klotz E, Mollet N, de Feyter PJ, Krestin GP. Stress myocardial perfusion: imaging with multidetector CT. Radiology 2014;270:25–46. [8] Petralia G, Bonello L, Viotti S, Preda L, d’Andrea G, Bellomi M. CT perfusion in oncology: how to do it. Cancer Imaging 2010;10:8–19. [9] Miles KA, Lee TY, Goh V, Klotz E, Cuenod C, Bisdas S, et al. Experimental Cancer Medicine Centre Imaging Network Group. Current status and guidelines for the assessment of tumour vascular support with dynamic contrast-enhanced computed tomography. Eur Radiol 2012;22:1430–41. [10] Imanishi Y, Fukui A, Niimi H, Itoh D, Nozaki K, Nakaji S, et al. Radiation-induced temporary hair loss as a radiation damage only occurring in patients who had the combination of MDCT and DSA. Eur Radiol 2005;15:41–6. [11] U.S. Food and Drug Administration. Safety investigation of CT brain perfusion scans; 2010 http://www.fda.gov/MedicalDevices/Safety/AlertsandNotices/ ucm185898.htm [12] International Commission on Radiological, Protection (ICRP). Publication, 103., The 2007 recommendations of the International Commission on Radiological, Protection. Ann ICRP 2007;37:2–4. [13] Hamada N, Fujimichi Y. Classification of radiation effects for dose limitation purposes: history, current situation and future prospects. J Radiat Res 2014;55:629–40. [14] Hamada N, Fujimichi Y, Iwasaki T, Fujii N, Furuhashi M, Kubo E, et al. Emerging issues in radiogenic cataracts and cardiovascular disease. J Radiat Res 2014, http://dx.doi.org/10.1093/jrr/rru036. [15] Preston DL, Ron E, Tokuoka S, Funamoto S, Nishi N, Soda M, et al. Solid cancer incidence in atomic bomb survivors: 1958–1998. Radiat Res 2007;168:1–64. [16] Ozasa K, Shimizu Y, Suyama A, Kasagi F, Soda M, Grant EJ, et al. Studies of the mortality of atomic bomb survivors, Report 14, 1950–2003: an overview of cancer and noncancer diseases. Radiat Res 2012 Mar;177(3):229–43. [17] United Nations Scientific Committee on the Effects of Atomic Radiation. Report to the general assembly, with scientific annexes. Volume I: Effects of ionizing radiation. New York: United Nations; 2006. [18] Shah DJ, Sachs RK, Wilson DJ. Radiation-induced cancer: a modern view. Br J Radiol 2012;85:e1166–73. [19] Durand DJ, Dixon RL, Morin RL. Ultilization strategies for cumulative dose estimates: a review and rational assessment. J Am Coll Radiol 2012;9:480–5. [20] Hendee WR, O’Connor MK. Radiation risks of medical imaging: separating fact from fantasy. Radiology 2012;264:312–21. [21] Shore RE. Radiation impacts on human health: certain, fuzzy, and unknown. Health Phys 2014;106:196–205. [22] Report of the United Nations Scientific Committee on the Effects of Atomic Radiation 2010. “Summary of low-dose radiation effects on health”. New York, NY: United Nations; 2011. [23] Hall EJ. Radiobiology for the radiologist. 5th ed. Philadelphia: Lippincott Williams & Wilkens; 2000. [24] International Commission on Radiological, Protection. Publication, 118., ICRP., Statement on tissue reactions/Early and late effects of radiation in normal



[25] [26]

[27]

[28]

[29]

[30]

[31]

[32]

[33] [34] [35]

[36]

[37]

[38]

[39]

[40] [41]

[42] [43] [44] [45] [46]

[47]

[48] [49]

[50]

[51] [52]


tissues and organs – threshold doses for tissue reactions in a radiation protection, context. Ann ICRP 2012;41(1/2). Geleijns J, Wondergem J. X-ray imaging and the skin: radiation biology, patient dosimetry and observed effects. Radiat Prot Dosim 2005;114:121–5. German Commission on Radiological Protection. Radiation-induced cataracts. http://www.ssk.de/SharedDocs/Beratungsergebnisse PDF/2009/Radiation Induced Cataracts.pdf?blob=publicationFile International Commission on Radiological, Protection. Publication, 60., 1990 recommendations of the International Commission on Radiological, Protection. Ann ICRP 1991;21:1–3. International Electrotechnical Commission. IEC 60601-2-44 Amend 1. Particular requirements for the basic safety and essential performance of X-ray equipment for computed tomography. Amendment 1. Geneva; 2012. Christner JA, Zavaletta VA, Eusemann CD, Walz-Flannigan AI, McCollough CH. Dose reduction in helical CT: dynamically adjustable z-axis X-ray beam collimation. Am J Roentgenol 2010;194:W49–55. International Commission on Radiological, Protection. Publication, 89., International Commission on Radiological, Protection. Basic anatomical and physiological data for use in radiological protection: reference, values. Ann ICRP 2002;32(3/4). Brix G, Lechel U, Veit R, Truckenbrodt R, Stamm G, Coppenrath EM, et al. Assessment of a theoretical formalism for dose estimation in CT: an anthropomorphic phantom study. Eur Radiol 2004;14:1275–84. Kalender WA, Schmidt B, Zankl M, Schmidt M. A PC program for estimating organ dose and effective dose values in computed tomography. Eur Radiol 1999;9:555–62. Stamm G, Nagel HD. CT-Expo-ein neuartiges Programm zur Dosisevaluierung in der CT. Fortschr Rontgenstr 2002;174:1570–6. Imaging Performance Assessment of CT-Scanners Group. ImPACT CT patient dosimetry calculator v. 1.0.3. http://www.impactscan.org Deak P, Smal Y, Kalender W. Gender-and age-specific conversation factors used to determine the effective dose from the dose-length product for multi-slice CT protocols. Radiology 2010;257:158–66. Committee to Assess Health Risks from Exposure to Low Levels of Ionizing Radiation. In: National Research Council. Health risks from exposure to low levels of ionizing radiation: BEIR VII Phase 2. Washington: The National Academies Press; 2006. Brix G, Berton M, Nekolla E, Lechel U, Schegerer A, Süselbeck T, et al. Cumulative radiation exposure and cancer risk of patients with ischemic heart diseases from diagnostic and therapeutic imaging procedures. Eur J Radiol 2013;82:1926–32. Shibuya K, Tsushima Y, Horisoko E, Noda SE, Taketomi-Takahashi A, Ohno T, et al. Blood flow change quantification in cervical cancer before and during radiation therapy using perfusion CT. J Radiat Res 2011;52:804–11. Li XS, Fan HX, Zhu HX, Song YL, Zhou CW. The value of perfusion CT in predicting the short-term response to synchronous radiochemotherapy for cervical squamous cancer. Eur Radiol 2012;22:617–24. Cullu N, Kantarci M, Ogul H, Adanur S, Okur A, Koc E, et al. Feasibility study of CT perfusion imaging for prostate carcinoma. Eur Radiol 2014;24:2236–44. Perisinakis K, Seimenis I, Tzedakis A, Papadakis AE, Damilakis J. The effect of head size/shape, miscentering, and bowtie filter on peak patient tissue doses from modern brain perfusion 256-slice CT: how can we minimize the risk for deterministic effects? Med Phys 2013;January (40):011911, http://dx.doi.org/10.1118/1.4773042. Statistisches Bundesamt. Statistisches Jahrbuch 2010 für die Bundesrepublik Deutschland. Stuttgart: Metzler-Poeschel; 2011. Robert Koch Institute, The Association of Population-based Cancer Registries in Germany, editors. Cancer in Germany 2009/2010, Berlin. 9th ed. 2014. International Agency for Research on Cancer. European Cancer Observatory (ECO). http://eco.iarc.fr/EUREG/AnalysisG.aspx International Commission on Radiological, Protection. Publication, 105., Radiological protection in, medicine. Ann ICRP 2007;37:6. International Atomic Energy Agency. Radiation protection and safety of radiation sources: International basic safety standards. General safety requirements Part 3. No. GSR Part 3 (Interim). Vienna: IAEA; 2011. Council of the European Union. Council directive 2013/59/Euratom of 5 December 2013 laying down basic safety standards for protection against the dangers arising from exposure to ionising radiation. American Heart, Association., Heart disease and stroke statistics – 2014 update: a report from the American Heart Association. Circulation 2014;129:e28–92. De Angelis R, Sant M, Coleman MP, Francisci S, Baili P, Pierannunzio D, et al. Cancer survival in Europe 1999–2007 by country and age: results of EUROCARE5 – a population-based study. Lancet Oncol 2014;15:23–34. Nichols M, Townsend N, Luengo-Fernandez R, Leal J, Gray A, Scarborough P, et al. European cardiovascular disease statistics 2012. Brussels/Sophia Antipolis: European Heart Network/European Society of Cardiology; 2012. Leiva-Salinas C, Provenzale JM, Wintermark M. Response to the 10 most frequently asked questions about perfusion CT. AJR 2011;196:53–60. American College of Radiology. ACR-ASNR-SPR practice guideline for the performance of computed tomography (CT) perfusion in neurologic imaging. http://www.acr.org/∼/media/ACR/Documents/PGTS/guidelines/CT Perfusion. pdf

[53] Kim SH, Kamaya A, Willmann JK. CT perfusion of the liver: principles and applications in oncology. Radiology 2014;272:322–44. [54] Cremer P, Hachamovitch R, Tamarappoo B. Clinical decision making with myocardial perfusion imaging in patients with known or suspected coronary artery disease. Semin Nucl Med 2014;44:330–429. [55] Jordan JD, Powers WJ. Cerebral autoregulation and acute ischemic stroke. Am J Hypertens 2012;25:946–50. [56] Brix G, Griebel J, Kiessling F, Wenz F. Tracer kinetic modelling of tumour angiogenesis based on dynamic contrast-enhanced CT and MRI measurements. Eur J Nucl Med Mol Imaging 2010;37(Suppl. 1):S30–51. [57] Brix G, Bahner ML, Hoffmann U, Horvath A, Schreiber W. Regional blood flow, capillary permeability, and compartment volumes: measurement with dynamic computed tomography – initial experience. Radiology 1999;210:269–76. [58] Hoffmann A, Zhu G, Wintermark M. Advanced neuroimaging in stroke patients: prediction of tissue fate and hemorrhagic transformation. Exper Rev Cardiovasc Ther 2012;10:515–24. [59] Brix G, Nekolla EN, Borowski M, Noßke D. Radiation risk and protection of patients in clinical SPECT/CT. Eur J Nucl Med Mol Imag 2014;41:S129–36. [60] Gowd BM, Heller GV, Parker MW. Stress-only SPECT myocardial perfusion imaging: a review. J Nucl Cardiol 2014;July. [61] Mettler FA, Buiberteau MJ. Essentials of nuclear medicine imaging. 6th ed. Philadelphia: Elsevier Saunders; 2012. [62] Miles KA. Measurement of tissue perfusion by dynamic computed tomography. Br J Radiol 1991;64:409–12. [63] Brix G, Zwick S, Griebel J, Fink C, Kiessling F. Estimation of tissue perfusion by dynamic contrast-enhanced imaging: simulation-based evaluation of the steepest-slope method. Eur Radiol 2010;20:2166–75. [64] Axel L. Cerebral blood flow determination by rapid-sequence computed tomography: a theoretical analysis. Radiology 1980;137:679–86. [65] Fieselmann A, Kowarschik M, Ganguly A, Hornegger J, Fahrig R. Deconvolutionbased CT and MR brain perfusion measurement: Theoretical model revisited and practical implementation details. Int J Biomed Imaging 2011;2011:467563, http://dx.doi.org/10.1155/2011/467563. [66] Johnson JA, Wilson TA. A model for capillary exchange. Am J Physiol 1966;210:1299–303. [67] StLawrence KS, Lee TY. An adiabatic approximation to the tissue homogeneity model for water exchange in the brain: I. Theoretical derivation. J Cereb Blood Flow Metab 1998;18:1365–77. [68] Tofts PS. Modeling tracer kinetics in dynamic Gd-DTPA MR imaging. J Magn Reson Imaging 1997;7:91–101. [69] Brix G, Zwick S, Kiessling F, Griebel J. Pharmacokinetic analysis of tissue microcirculation using nested models: multimodel inference and parameter identifiability. Med Phys 2009;36:2923–33. [70] Brix G, Ravesh MS, Zwick S, Griebel J, Delorme S. On impulse response functions computed from dynamic contrast-enhanced image data by algebraic deconvolution and compartmental modeling. Phys Med 2012;28:119–28. [71] Sourbron SP, Buckley DL. Classic models for dynamic contrast-enhanced MRI. NMR Biomed 2013;26:1004–27. [72] Ingrisch M, Sourbron S. Tracer-kinetic modeling of dynamic contrast-enhanced MRI and CT: a primer. J Pharmacokinet Pharmacodyn 2013;40:281–300. [73] Kalender WA, Deak P, Kellermeier M, van Straten M, Vollmar SV. Applicationand patient size-dependent optimization of X-ray spectra for CT. Med Phys 2009;36:993–1007. [74] Kalender WA. Dose in X-ray computed tomography. Phys Med Biol 2014;59:R129–50. [75] Gudjónsdóttir J, Ween B, Olsen DR. Optimal use of AEC in CT: a literature review. Radiol Technol 2010;81:309–17. [76] Wang J, Duan X, Christner JA, Leng S, Yu L, McCollough CH. Radiation dose reduction to the breast in thoracic CT: comparison of bismuth shielding, organbased tube current modulation, and use of a globally decreased tube current. Med Phys 2011;38:6084–92. [77] Beister M, Kolditz D, Kalender WA. Iterative reconstruction methods in X-ray CT. Phys Med 2012;28:94–108. [78] Tao Y, Chen GH, Hacker TA, Raval AN, Van Lysel MS, Speidel MA. Low dose dynamic CT myocardial perfusion imaging using a statistical iterative reconstruction method. Med Phys 2014;41(July):071914, http://dx.doi.org/10.1118/1.4884023. [79] Lauzier PT, Chen GH. Characterization of statistical prior image constrained compressed sensing. I. Applications to time-resolved contrast-enhanced CT. Med Phys 2012;39:5930–48. [80] Brix G, Lechel U, Petersheim M, Krissak R, Fink C. Dynamic contrast-enhanced CT studies: balancing patient exposure and image noise. Invest Radiol 2011;46:64–70. [81] Chung JJ, Cho ES, Kang SM, Yu JS, Kim DJ, Kim JH. Usefulness of a lead shielding device for reducing the radiation dose to tissues outside the primary beams during CT. Radiol Med 2014 [Epub ahead of print]. [82] Geleijns J, Salvadó Artells M, Veldkamp WJ, López Tortosa M, Calzado Cantera A. Quantitative assessment of selective in-plane shielding of tissues in computed tomography through evaluation of absorbed dose and image quality. Eur Radiol 2006;16:2334–40. [83] Colletti PM, Micheli OA, Lee KH. To shield or not to shield: application of bismuth breast shields. AJR 2013;200:503–7.


Measurement of tissue perfusion by dynamic computed tomography.

Measurement of tissue perfusion by dynamic computed tomography.

Abdominal perfusion computed tomography.

Radiological protection in computed tomography and cone beam computed tomography.

computed tomography imaging.

Perfusion computed tomography in renal cell carcinoma.

Perfusion patterns of ischemic stroke on computed tomography perfusion.

Update on Computed Tomography Myocardial Perfusion Imaging.

Development and validation of a segmentation-free polyenergetic algorithm for dynamic perfusion computed tomography.

Evaluation of static and dynamic perfusion cardiac computed tomography for quantitation and classification tasks.

computed tomography images in pulmonary embolism.

Computed tomography in planning radiation therapy.

Computed tomography in radiation therapy treatment planning.

Low contrast medium and radiation dose for hepatic computed tomography perfusion of rabbit VX2 tumor.

Clinical application of perfusion computed tomography in neurosurgery.

Radiation dose optimization and thoracic computed tomography.

Computed tomography perfusion imaging for the assessment of brain tumors.

Colour perfusion imaging: a new application of computed tomography.

Computed tomography number changes observed during computed tomography-guided radiation therapy for head and neck cancer.

Perfusion computed tomography: an imaging biomarker for brain tumors' grading.

Correlation of quantitative dual-energy computed tomography iodine maps and abdominal computed tomography perfusion measurements: are single-acquisition dual-energy computed tomography iodine maps more than a reduced-dose surrogate of conventional computed tomography perfusion?

Blood-brain barrier permeability imaging using perfusion computed tomography.

Perfusion computed tomography to assist decision making for stroke thrombolysis.

Stroke thrombolysis at 5.5 hours based on computed tomography perfusion.