Méthodes de réduction de la dose en tomodensitométrie

Dose and Radiation Risk in CT

Today, in industrialized countries, the radiation dose level from medical exposure is in the same range as the annual natural background radiation of 3.1 mSv. Radiation dose in medical imaging is an important topic of discussion. In the following section, in-depth information is provided about how radiation dose during a CT scan is estimated. Furthermore, the factors that affect it are discussed, including the difficulties in analyzing the real risk that can be attributed to CT scans and some interesting comparisons with environmental influences. 

During a CT scan, cross-sections (slices) of the body are irradiated. Nevertheless, the X-ray dose delivered to the body is not exactly confined to the user-defined slices, but extends outside this area due to scattering of the radiation (Figure 1).

The scattering of the X-rays must be included when calculating the absorbed dose D. The Computed Tomography Dose Index (CTDI) is the sum of the absorbed dose in the slices and the contributions outside (Figures 2 and 3), normalized to the nominal slice thickness.

Mathematically, the CTDI is calculated as the integral of the absorbed dose along the z-axis, divided by the nominal slice thickness.
CTDI is the measure of the dose deposited in a single axial slice of the patient. The unit used to measure it is the mGy (1 mGy = 1/1000 Gy).
In practice, the integration limits cannot be extended to infinity. E.g. CTDI as defined by the FDA (Food and Drug Administration) in the United States requires an integration length of 7 nominal slice thicknesses on either side of the irradiated slice¹. The more common definition today, CTDI¹⁰⁰, requires an integration range of 50 mm on either side of the irradiated slice. This is more practical, since most ionization chambers used to measure CTDI are 100 mm long. The ionization chambers are placed in the center and the periphery of Perspex® dummies of 16 cm diameter for the head and 32 cm diameter for the body (Figure 4).

There are different ways to calculate the CTDI. One of them is to consider the differences between the absorbed dose in the periphery and in the center of the patient’s body by a weighted sum of the central and peripheral CTDI values.

The resulting formula for the weighted CTDI (CTDI_w) that takes into account this difference is:
CTDI_W=1/3CTDI^A₁₀₀+2/3CTDI^B₁₀₀

¹FDA October 20, 2006: Guidance for Industry, FDA Staff, and Third Parties – Provision for Alternate Measure of the Computed Tomography Dose Index (CTDI) to Assure Compliance with the Dose Information Requirements of the Federal Performance Standard for Computed Tomography

Volume CT scans include many sequential slices during a spiral scan. For this reason, the velocity with which the table moves must be considered: If the table moves slowly, the X-ray beam profiles will overlap (Figure 5). For a spiral scan, pitch is defined as the longitudinal distance in mm that the table travels during one revolution of the X-ray tube divided by the nominal irradiated width of the detector projected to the isocenter of the scanner.

For a spiral examination, the CTDI_vol is:

CTDI_vol = CTDI_w 1/pitch

If the pitch is smaller than 1, the X-ray beam profiles overlap and the absorbed dose increases. If the pitch is larger than 1, the X-ray beam profiles do not overlap, there are gaps in the acquisition and the absorbed dose decreases. This is valid for both single-detector and multidetector row CT.

The expected CTDI_vol is displayed on the user interface of the CT scanner prior to each scan. The operator can therefore easily observe on the screen the absorbed dose according to the parameters chosen for the scan (Figure 6).

In order to calculate the total absorbed dose for a complete CT examination, the range that is being examined must be taken into account (Figure 7).

The dose length product (DLP) is the product of CTDI_vol and the examination range: DLP = CTDI_vol · L It is measured in mGy cm. Both CTDI_vol and DLP for each CT examination are stored with the patient protocol and are therefore readily available.

Another aspect to be considered is that the absorbed dose is also related to the size of the patient. If a patient is smaller than the 32-cm Perspex® phantom used to determine the body CTDI, the actual absorbed dose will be higher. If the patient is bigger, the actual absorbed dose will be lower.

If the patient’s shape/cross-section is similar to that of the CTDI phantom, CTDI_vol can be used as an estimate for absorbed patient dose.

The effective dose in CT takes into account the direct and scattered radiation for all organs in the scan volume. It cannot be calculated exactly for each patient, but a close estimate can be obtained by means of Monte Carlo simulations, assuming an idealized “average” patient. It is based on a mathematical adult hermaphrodite phantom of the kind used for Monte Carlo simulations of effective doses by the UK National Radiological Protection Board (NRBP) in 1989.

The formula for the effective dose E is: E = Σ D_org · w_org

The effective dose in CT is therefore a measure of the mean radiation burden based on a patient group, not a measure of the radiation burden of an individual patient, who normally deviates from the idealized “average” patient).

The effective dose is the sum of the doses for all organs, multiplied by the respective tissue weighting factors. For different scan ranges, the effective dose E can be calculated approximately from the dose length product (DLP)¹:

E = DLP · f

The mean weighting factor f (average between male and female models) is used for different regions of the human body:

1. Head: f = 0.0021 mSv/(mGy · cm)
2. Neck: f = 0.0059 mSv/(mGy · cm)
3. Thorax: f = 0.014 mSv/(mGy · cm)
4. Abdomen and Pelvis: f = 0.015 mSv/(mGy · cm)

Table 1 shows typical examples of the effective dose for different CT routines.

¹Jessen KA et al. EUR 16262: European Guidelines on Quality Criteria for Computed Tomography. Office for Official Publications of the European Communities, Luxembourg, 2000.

The effective doses typically used during CT routines (e. g. head 1.9 mSv, thorax 3.4 mSv, abdomen 4.9 mSv) are far below the thresholds that are commonly associated with deterministic damage (Table 2). 

However, the risk of stochastic damage after one CT scan remains uncertain. There are only a few assumptions and models to quantify this risk.
The most important study that addresses this issue was conducted on 105,000 radiation victims in Hiroshima and Nagasaki, of which 35,000 received radiation doses between 5 and 200 mSv.¹ Unfortunately, this study revealed a high statistical uncertainty in the low dose range that applies to CT scanning. Furthermore, uncertainties remain about the shape of the dose-response, both for cancer and for non-cancer diseases, below about 100 mSv.² The assumption today is a linear relationship between the radiation dose and the additional cancer risk with no dose threshold (linear no-threshold model, or LNT) and that risk depends strongly on the age at the time of irradiation (the younger the child, the higher the potential risk).

A recent publication by Brenner et al estimated the lifetime risk of death from cancer attributable to a CT scan as shown in Figure 9A and 9B. ³
The International Commission on Radiological Protection (ICRP) of 1990 assumed an excess lifetime cancer mortality risk of about 5% per Sv. Based on this assumption, a CT examination with 10 mSv may increase cancer mortality risk by about 0.05%. This value is in reasonable agreement with Brenner’s assumptions (Figures 9A and 9B). However, this risk has to be framed appropriately. According to Smith, the average cancer mortality risk in Western society is about 25%. After a CT examination with 10 mSv, it is increased by only 0.05% (25.05%). This is the same increase of mortality risk as living in Central London for 450 days (death caused by air pollution) or living in the same apartment with a smoker for 540 days.⁴
The estimated lifetime risk of death from various sources is shown in Table 3.

Therefore, if clinically indicated, the benefit of a CT examination far outweighs the additional radiation risk for the patient. Nevertheless, Siemens’ ultimate goal is to adhere to the ALARA (As Low As Reasonably Achievable) principle, i.e. to use the reasonably achievable dose to obtain the required diagnostic quality images.

¹ Preston DL et al. Solid cancer incidence in atomic bomb survivors: 1958–1998. Radiat Res. 2007 Jul;168(1):1-64.
² Muirhead CR. Studies on the Hiroshima and Nagasaki survivors, and their use in estimating radiation risks. Radiat Prot Dosimetry. 2003;104(4):331-5.
³ Brenner DJ et al. Computed tomography – an increasing source of radiation exposure. N Engl J Med. 2007 Nov 29;357(22):2277-84.
⁴ Smith JT. Are passive smoking, air pollution and obesity a greater mortality risk than major radiation incidents? BMC Public Health. 2007 Apr 3;7:49.

Technological Advances in CT Dose Reduction

Siemens invests heavily in implementing all of the dose reduction methods possible in CT today and was the first to implement many dose-saving features into clinical routine.

To maintain its leading position and to help improve health care for patients, Siemens cooperates closely with experts from around the globe in universities, public clinics and private radiology centers to bring research developments into practical, everyday clinical routine.

In addition to the newest technology, dose reduction in CT requires training to develop familiarity with dose reduction methods and factors. That’s why Siemens therefore takes all possible steps to make the dose savings products as intuitive as possible to physicians and technologists and also offers on-going seminars and resources related to dose reduction.

The following chapters provide brief descriptions of our dose reduction products and algorithms.

The most efficient way to reduce radiation dose in CT is by adaptating the scan parameters to the patient’s anatomy. Centering the patient correctly, using the right protocols and adjusting the X-ray tube output to the patient’s size and shape help to minimize radiation exposure.

Some users, however, may not fully be aware of how parameters should be modified to adjust radiation dose levels for different patients. For example, they may not realize that – when scanning an area where the patient’s diameter has decreased by only 4 cm – the tube output can be reduced by a factor of two while still maintaining adequate image quality. Hence, all modern Siemens CT scanners provide control mechanisms that automatically adjust the radiation dose level to the patient’s anatomy.

Siemens CARE Dose4D automatically adapts the tube current to the size and shape of the patient, achieving optimal tube current modulation in two ways^{1, 2}. First, tube current is varied on the basis of a topogram, by comparing the actual patient to a “standard-sized” patient. As might be expected, tube current is increased for larger patients and reduced for smaller patients.

Differences in attenuation in distinct body regions are taken into account. For example, in an adult patient, 140 mAs might be needed in the shoulder region, whereas 55 mAs would be sufficient in the thorax, 110 mAs in the abdomen and 130 mAs in the pelvis.

Second, real-time angular dose modulation measures the actual attenuation in the patient during the scan and adjusts tube current accordingly – not only for different body regions, but also for different angles during rotation. This is particularly important for efficiently reducing dose in the shoulder and pelvic region, where the lateral attenuation is much higher than the anterior-posterior attenuation. Figure 1 demonstrates the working principle of CARE Dose4D.

Figure 2 shows a clinical example of optimized radiation dose for various anatomic regions obtained with the use of CARE Dose4D.

Clinical experience has revealed that there is no linear relationship between optimal tube current and patient attenuation. Larger patients clearly need a higher dose than average-sized patients, but they also have more body fat, which increases tissue contrast. Smaller patients need a lower dose than average-sized patients, but they have less fat and less tissue contrast, which would result in noisy images if the dose were too low. Therefore, during real-time dose modulation, CARE Dose4D reduces radiation dose less than might be expected for smaller patients, while increasing the dose less than might be expected for larger patients. This maintains good diagnostic image quality while achieving an optimal radiation dose (Figure 3). When modulating the tube current in x-, y- and z-direction with CARE Dose4D the radiation dose may be reduced.

¹ Greess H et al. Dose reduction in subsecond multislice spiral CT examination of children by online tube current modulation. Eur Radiol. 2004 Jun;14(6):995-9.

² Alibek S et al. Dose reduction in pediatric computed tomography with automated exposure control. Acad Radiol. 2011 Jun;18(6):690-3.
 

With this method, the radiation dose is modulated during the complete spiral CT scan by using information from the patient’s ECG. The tube current is maintained at 100% of the desired level only during a predefined “phase of interest” of the patient’s cardiac cycle. During the rest of the time the current can be reduced, thus potentially reducing the mean radiation dose (Figure 4).¹

ECG-controlled dose modulation is based on continuous monitoring of the ECG and an algorithm that predicts when the desired ECG phase will start by calculating the mean durations of the preceding cardiac cycles. Older ECG-pulsing methods reach their limitations with arrhythmia patient scans that cannot be predicted by simple averaging. Recently, more versatile ECG-pulsing algorithms have been introduced that react flexibly to arrhythmia and ectopic beats and have the potential to considerably enhance the clinical application spectrum of ECG-controlled dose modulation.

¹ Jakobs TF et al. Multislice helical CT of the heart with retrospective ECG gating: reduction of radiation exposure by ECG-controlled tube current modulation. Eur Radiol. 2002 May;12(5):1081-6.
 

Prospective ECG-triggering combined with “step-and-shoot” acquisition of axial slices is a very dose-efficient way of ECG-synchronized scanning. In fact, only the very minimum of scan data needed for image reconstruction is acquired during the previously selected heart phase.

The patient’s ECG signal is monitored during examination, and axial scans are started with a pre-defined temporal offset relative to the R-waves. With conventional approaches, the method reaches its limitations in patients with severe arrhythmia, since ECG-triggered axial scanning depends on a reliable prediction of the patient’s next cardiac cycle by using the mean length of the preceding cardiac cycles

With Adaptive Cardio Sequence, a more refined analysis of the patient’s ECG is performed. Irregularities are reliably detected. In case of an ectopic beat, the scan can be either skipped if the ectopic beat happens earlier than the predicted scan start, thus saving unnecessary dose, or repeated at the same position. Hence, the application spectrum of ECG-triggered sequential scanning is extended to patients with high and irregular heart rates (Figures 5 and 6).
 

In spiral CT, it is routine to do an extra half-rotation of the gantry before and after each scan, fully irradiating the detector throughout, even though only part of the acquired data is necessary for image reconstruction. This problem is typical for spiral CT and commonly referred to as “over-ranging” (Figure 7).

The Adaptive Dose Shield, a technology based on precise, fast and independent movement of both collimator blades, limits this over-ranging. The pre-patient collimator asymmetrically opens and closes at the beginning and end of each spiral scan, temporarily blocking those parts of the X-ray beam that are not used for image reconstruction. As a result, only the targeted tissue is irradiated (Figure 8).

Measurements at the Institute of Medical Physics, University Erlangen-Nuernberg, Germany,¹ and at the Clinical Innovation Center, Mayo Clinic, Rochester, Minnesota, USA,² have demonstrated significant dose reductions, depending on the scanned range, without affecting image quality (Figure 9).

¹ Deak PD et al. Effects of adaptive section collimation on patient radiation dose in multisection spiral CT. Radiology. 2009 Jul;252(1):140-7.
² Christner JA et al. Dose reduction in helical CT: dynamically adjustable z-axis X-ray beam collimation. AJR Am J Roentgenol. 2010 Jan;194(1):W49-55.

³ This study evaluated dose reduction on a commercial CT scanner with and without adaptive section collimation to reduce pre-spiral and post-spiral radiaton. Measurements were made with thermoluminescent dosimeters in CT dose index phantoms and in an Alderson-Rando phantom for spiral cardiac and chest CT protocols and were compared with the Monte Carlo simulated dose profiles.

Dual Source CT (DSCT) provides a way to scan the heart within one heartbeat without using an area detector that covers the entire heart volume. With a single source CT (Figure 10 left), the spiral pitch is limited to values below 2.0 to enable gapless volume coverage along the z-axis. If the pitch is increased, sampling gaps occur that hamper the reconstruction of images with well-defined narrow slice sensitivity profiles and without excessive image artifacts. With DSCT systems, however, data acquired with the second measurement system a quarter rotation later can be used to fill these gaps (Figure 10 right). In this way, the pitch can be increased up to 3.4 in a scan field of view (SFOV) that is covered by both detectors. Since no redundant data are acquired due to the high pitch, a quarter rotation of data per measurement system is used for image reconstruction, and each of the individual axial images has a temporal resolution of a quarter of the rotation time.

The SOMATOM Definition Flash offers 38.4 mm detector z-coverage and 0.28 s gantry rotation time. At a pitch of 3.4, the table feed is 450 mm/s, which is sufficient to cover the heart (12 cm) in about 0.27 s. The scan is triggered and starts at a user-selectable phase of the patient’s cardiac cycle. Each of the images has a temporal resolution of 75 ms, and the phase of images adjacent in the z-direction is slightly shifted (Figure 11).

Since no overlapping data are acquired, the radiation dose of this new mode is very low and even below the dose values of ECG-triggered sequential scanning. Initial publications have demonstrated that reliable coronary CT angiography (CTA) is feasible at radiation dose values below 1 mSv.^1,2

Figure 12 shows images reconstructed in this mode with an acquisition time of 250 ms, a temporal resolution of 75 ms, 100 kV and a resulting effective dose of 0.8 mSv.

The first scientific papers to be published on the SOMATOM Definition Flash demonstrated effective radiation doses of 0.88–0.9 mSv for routine coronary CTA.^{1, 2}

¹ Achenbach S et al. Coronary computed tomography angiography with a consistent dose below 1 mSv using prospectively electrocardiogram-triggered high-pitch spiral acquisition. Eur Heart J. 2010 Feb;31(3):340-6.
² Leschka S et al. Diagnostic accuracy of high-pitch dual-source CT for the assessment of coronary stenoses: first experience. Eur Radiol. 2009 Dec;19(12):2896-903.

According to recently modified tissue weighting factors (recommendations of the International Commission on Radiological Protection of 2007, ICRP103), the female breast is more radiosensitive than previously assumed. In any CT examination of the thorax, the breast – even without being the object of interest – is irradiated and should therefore be especially protected. Siemens X-CARE, an organ-based dose modulation mode, can selectively limit the radiation exposure of sensitive organs. When using this mode, radiation intensity is reduced when the patient is irradiated from the front (Figure 13).

With this method, the radiation exposure of the breast or the eyes is reduced (Figure 14).

Iterative reconstruction has been a topic of interest in computed tomography for decades. In the 1980s, researchers were already attempting to improve image quality with several iterative reconstruction loops. Image imperfections were identified by comparing reconstructed images with the measured data in the raw date space. In the next reconstruction loop the images were improved based on this comparison.

But especially transferring the data from a reconstructed image (the so-called image space) back into the raw data space is time consuming. In the past, long calculation times prevented the use of iterative imaging in clinical routine, as this could take up to several hours for large datasets.

IRIS – Iterative Reconstruction in Image Space – is a Siemens unique approach to iterative reconstruction. It is mathematically proven that noise reduction can be completely separated from artifact reduction. Noise reduction can be achieved with iterations in image space only, without transferring the images back into raw data space. Therefore IRIS is performing iterative loops in image space (Fig. 15). When acquiring CT examinations with lower dose the image noise increases. With IRIS the noise can be cleaned up (Fig. 16).^{1, 2}

¹ Bulla S et al. Reducing the radiation dose for low-dose CT of the paranasal sinuses using iterative reconstruction: Feasibility and image quality. Eur J Radiol. 2011 Jun 8. [Epub ahead of print]

² Pontana F et al. Chest computed tomography using iterative reconstruction vs filtered back projection (Part 2): image quality of low-dose CT examinations in 80 patients. Eur Radiol. 2011 Mar;21(3):636-43.

SAFIRE is the first raw data-based iterative reconstruction from Siemens. Noise reduction, and hence dose reduction, is achieved when performing iterative loops in image space. In addition SAFIRE transfers the data back into the raw data space and also performs iterative loops there (Figure 17). Raw data are visualized in the so-called sinogram, hence the name SAFIRE, Sinogram Affirmed Iterative Reconstruction. Iterations in raw data space improve image quality with regards to artifacts, contrast, and sharpness. SAFIRE has the potential to reduce dose by up to 60% or improve image quality.¹

The iterations in raw data space are computationally intensive. The new image reconstruction system (IRS) is especially designed to support iterative reconstruction algorithms. The image reconstruction speed of up to 20 images per second for the SOMATOM Definition Family brings the performance to a level comparable to the processing of conventional CT examinations. With this, SAFIRE can even  be used in acute care or other time-sensitive clinical applications.

SAFIRE entered the U.S. market in June 2013. At the time SAFIRE entered the market, the quantitative dose reduction potential provided by SAFIRE was unique in the market.

So SAFIRE  reduces noise or improves image quality, and can be used in the clinical routine – for significant dose reduction and high performance (Figure 18)².

¹Moscariello A et al. Coronary CT angiography: image quality, diagnostic accuracy, and potential for radiation dose reduction using a novel iterative image reconstruction technique – comparison with traditional filtered back projection. Eur Radiol. 2011 Oct;21(10):2130-8.

²In clinical practice, the use of SAFIRE may reduce CT patient dose depending on the clinical task, patient size, anatomical location, and clinical practice. A consultation with a radiologist and a physicist should be made to determine the appropriate dose to obtain diagnostic image quality for the particular clinical task. The following test method was used to determine a 54 to 60% dose reduction when using the SAFIRE reconstruction software. Noise, CT numbers, homogeneity, low-contrast resolution and high contrast resolution were assessed in a Gammex 438 phantom. Low dose data reconstructed with SAFIRE showed the same image quality compared to full dose data based on this test. Data on file.

Conventional dose modulation approaches control only the X-ray tube current while the X-ray tube voltage (the kV setting) is left untouched. Yet, there is great potential for dose reduction by adapting the kV setting.

CARE kV is a fully automated feature that adjusts the tube voltage tailored to the individual patient, the system capabilities and the clinical task. When changing the tube voltage the tube current needs to be adapted as well. This is necessary to maintain a constant contrast-to-noise ratio (CNR). CNR is the technical way to define image quality. So in combination CARE kV and CARE Dose4D allow the patient-specific adaption of both parameters, tube voltage and tube current.

Additionally, an iodine contrast agent is often administered to improve contrast and thus the visibility of organ structures in CT images (particularly in CT angiography). The contrast is best with lowered X-ray tube voltage, since the low energy X-rays are better absorbed by the dense iodine than by the surrounding tissue. So in CT Angiography examinations dose can be
significantly reduced by choosing 80 kV or 100 kV instead of 120 kV (Figure 19)¹.

For larger patients, though, who have a higher X-ray attenuation, the output of the X-ray tube at lower kV settings may not be sufficient to produce the required contrast-to-noise ratio. For these patients, higher X-ray tube voltages will have to be selected.
In a busy environment, technicians and reading physicians often have insufficient time to measure the attenuation of each patient. Automatic tools that define the optimal combination of voltage and current for each patient according to the patient’s topogram and the selected examination protocol are therefore necessary (Figure 20).

¹ Winklehner A et al. Automated attenuation-based tube potential selection for thoracoabdominal computed tomography angiography: improved dose effectiveness. Invest Radiol. 2011 Dec;46(12):767-73.

Choosing the parameters and setting for CT examinations of pediatric patients can be challenging. And special attention needs to be paid as children are more sensitive to radiation than adults. With CARE Child a full package of tools is provided helping to set the right adjustments. Dedicated pediatric standard protocols are shipped with the system for a wide range of clinical questions. For CARE Dose4D, Siemens automated tube current modulation, special adjustment curves for children are implemented for different clinical tasks. These curves are designed for the imaging of smaller bodies and hence, smaller structures. While CARE Dose4D provides an optimization of the tube current, CARE kV proposes the tube voltage setting automatically, including the possibility of scanning with 70 kV. The 70 kV mode is a Siemens technology also set up for the imaging of the youngest patients. Equipped with these tools pediatric patients can obtain CT examinations at a low dose with excellent image quality.

Conventional solid-state detectors consist of a scintillator layer that converts the incoming X-rays into visible light, a photodiode array that converts the visible light into an electric current and an analog-to-digital converter (ADC) which digitizes the signal on a separate electronic board. The number of electronic components and relatively long conducting paths increase power consumption, and add to the electronic noise produced by the detector.
The Stellar Detector is the first fully integrated detector. The photodiode and the ADC have been integrated next to each other, reducing the path of the signal (Fig. 21). The transfer of the digitized signal is done without any losses and the electronic noise produced by the detector is reduced by a factor of two (TrueSignal Technology).

In clinical CT, the attenuation of the measured object varies dramatically and so do the signal levels at the detector. The dynamic range describes the range of the input signal levels that can be reliably measured simultaneously without saturation. HiDynamics has an exceptionally high dynamic range of 102 dB. Combined with the noise reduction provided by TrueSignal Technology, Stellar Detectors can measure smaller signals over a wider dynamic range which directly enhances CT image quality especially for applications with extremely low signal levels. Such extremely low signal levels play an important role when scanning large patients and in low dose scans, as well as in the low-kV datasets of Dual Energy examinations.

Pediatric Computed Tomography
Radiographic examinations are used much less frequently for children than for adults, because their organism is still developing and because children seldom understand the cooperation (such as breath-holds, etc.) required of them. Nevertheless, computed tomography is of great importance for imaging of pediatric patients, especially for complex lung imaging, for imaging of congenital malformations, and in intensive care. As a consequence, the ALARA principle (As Low As Reasonably Achievable) is of particular importance in pediatrics. It calls for always selecting the dose that is as low as reasonably achievable, yet sufficient for a reliable diagnosis.