Dose and image quality of cone-beam computed tomography as compared with conventional multislice computed tomography in abdominal imaging.

ORIGINAL ARTICLE

Dose and Image Quality of Cone-Beam Computed Tomography as Compared With Conventional Multislice Computed Tomography in Abdominal Imaging Alexander A. Schegerer, PhD,* Ursula Lechel, MPhys,* Manuel Ritter, PhD,Þ Gerald Weisser, PhD,þ Christian Fink, PhD,§ and Gunnar Brix, PhD* Objectives: Recent technical developments have facilitated the application of cone-beam computed tomography (CBCT) for interventional and intraoperative imaging. The aim of this study was to compare the radiation doses and image quality in CBCT with those of conventional multislice spiral computed tomography (MSCT) for abdominal and genitourinary imaging. Methods: Different CBCT and MSCT protocols for imaging soft tissues and hard-contrast objects at different dose levels were investigated in this study. Local skin and organ doses were measured with thermoluminescent dosimeters placed in an anthropomorphic phantom. Moreover, the contrast-to-noise ratio, the noise-power spectrum, and the high-contrast resolution derived from the modulation transfer function were determined in a phantom with the same absorption properties as those of anthropomorphic phantom. Results: The effective dose of the examined abdominal/genitourinary CBCT protocols ranged between 0.35 mSv and 18.1 mSv. As compared with MSCT, the local skin dose of CBCT examinations could locally reach much higher doses up to 190 mGy. The effective dose necessary to realize the same contrast-to-noise ratio with CBCT and MSCT depended on the MSCT convolution kernel: the MSCT dose was smaller than the corresponding CBCT dose for a soft kernel but higher than that for a hard kernel. The noise-power spectrum of the CBCT images at tube voltages of 85/90 kV(p) is at least half of that of images measured at 103/115 kV(p) at any arbitrarily chosen spatial frequency. Although the pixel size and slice thickness of CBCT were half of those of the MSCT images, high-contrast resolution was inferior to the MSCT images reconstructed with a hard convolution kernel. Conclusions: As compared with MSCT using a medium-hard convolution kernel, CBCT produces images at medium noise levels and, simultaneously, medium spatial resolution at approximately the same dose. It is well suited for visualizing hard-contrast objects in the abdomen with relatively low image noise and patient dose. For the detection of low-contrast objects at standard tube voltages of approximately 120 kV(p), however, MSCT should be preferred. Key Words: cone-beam CT, multislice spiral CT, radiation exposure, image quality, abdominal imaging (Invest Radiol 2014;49: 675Y684)

T

he application of cone-beam computed tomography (CBCT) for image-guided interventions and surgical procedures outside the radiology department has strongly increased. Cone-beam computed

Received for publication February 13, 2014; and accepted for publication, after revision, March 25, 2014. From the *Department of Medical and Occupational Radiation Protection, Federal Office for Radiation Protection, Neuherberg; †Department of Urology, ‡Institute of Clinical Radiology and Nuclear Medicine, Medical Centre Mannheim, Mannheim; and §Abteilung fu¨r Radiologie, Allgemeines Krankenhaus Celle, Celle, Germany. Johannes Berndt and Markus Oechsner, Klinikum rechts der Isar, Munich, Germany, supported the measurements at the multislice spiral computed tomography system. Conflicts of interest and sources of funding: none declared. Reprints: Alexander A. Schegerer, PhD, Bundesamt fu¨r Strahlenschutz, Abteilung fu¨r medizinischen und beruflichen Strahlenschutz, Ingolsta¨dter LandstraQe 1, 85764 Neuherberg, Germany. E-mail: [email protected]. Copyright * 2014 by Lippincott Williams & Wilkins ISSN: 0020-9996/14/4910Y0675

Investigative Radiology

& Volume 49, Number 10, October 2014

tomography has been established in many different clinical fields, for example, neurology,1Y3 otolaryngology,4 cardiology,5,6 abdominal angiography,7,8 musculoskeletal imaging,9,10 and, recently, urology. The promising CBCT technology is a refinement of conventional CarmYbased x-ray systems with pulsed radiation and improved image processing algorithms. Its most characteristic feature is an indirectly converting cesium iodide flat panel detector rotating around the patient to get a set of projection data that can be used for computed tomographic (CT) image reconstruction. The clinical relevance of CBCT systems results from their flexible use for projection radiography, fluoroscopy, digital subtraction angiography, and, in particular, cross-sectional, multiplanar imaging in multiple viewing planes in a single-patient setup. The latter feature is of particular relevance for interventional and surgical procedures because the improved delineation of soft tissue lesions (eg, tumors) and that of small hard-contrast objects (eg, ureteral stones) on CT images, in contrast to conventional projection radiography or fluoroscopy, reduce not only the time required for therapeutic procedures but also therapy-associated complications.11Y13 As compared with postoperative conventional multislice CT (MSCT) performed in the radiology department, the use of a CBCT system in the operating room avoids a putative second intervention requiring further anesthesia and substantially reduces the problem of postoperative anatomical changes and different positioning of the patient. Accordingly, intraoperative CBCT improves workflow and reduces costs. In contrast to a conventional MSCT system, the intrinsic noise of (currently used) CBCT detectors reaches a minimum at tube voltages of approximately 85 kV(p).14 Such low tube voltages can result both in increased patient exposure to radiation for constant image quality and in more disturbing beam hardening artifacts, for example, caused by an administered contrast agent or by a metallic catheter piston. Furthermore, the broad cone-beam geometry can yield image distortions.15 The large cone angle also increases scattered radiation as well as noise and, which, in turn, reduces the detectability of lowcontrast objects. To date, there are only few reports in the literature on radiation doses and image quality for CBCT. The aim of the present phantom study was to compare patient exposure and image quality of various CBCT and MSCT protocols for abdominal and genitourinary imaging.

MATERIALS AND METHODS CT Systems and Protocols Measurements were performed using a CBCT system (Artis Zee Ceiling Dyna-CT; Siemens Healthcare, Forchheim, Germany) installed in an operating room of a urological department of a university medical center. The system is equipped with a newly developed full carbon intervention table and a rotable flat panel detector with a size of 30 cm 40 cm consisting of 2000 2000 detector elements. The ceiling-mounted installation is space-saving in contrast to its predecessors. The x-ray tube and detector rotate by approximately 200 degrees around the patient to get a full set of projection www.investigativeradiology.com

Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

675


Schegerer et al


TABLE 1. Protocols Used at the CBCT System for Abdominal and Genitourinary Imaging

No. Protocol CBCT-I CBCT-II CBCT-III CBCT-IV

Scan Time, s

Dose/ Projection, KGy

Pulse Width, mm*

Angulation Step, degree

No. Projections, degree

Voltage, kV(p)†

DAP, cGyIcm2‡

Pixel Size, mm§

Slice Thickness, mm||

5 8 8 20

0.08 0.08 0.36 1.2

3.4 3.5 3.9 10.1

1.5 0.5 0.4 0.4

133 397 397 496

115 85 90 103

165 796 4437 10,176

0.49 0.49 0.49 0.49

0.49 0.49 0.49 0.49

*x-Ray pulse width per projection. †Tube voltage. ‡DAP averaged over 5 scans. §Pixel size of the reconstructed image. ||Slice thickness of the reconstructed image. CBCT indicates cone-beam computed tomography; DAP, dose area product.

data for the reconstruction of cross-sectional images. Threedimensional image reconstruction and image postprocessing were performed with a dedicated workstation and a three-dimensional reconstruction software (MMWP syngo VE 40A; Siemens Healthcare, Forchheim, Germany). By default, images with a matrix size of 512 512 were reconstructed. The system is calibrated to Hounsfield units. Four different protocols were investigated (Table 1). In previous clinical trials, the parameter settings of protocols CBCT-I and CBCT-II were adapted for the needs in the urological operating room, that is, to localize stones and improve imaging of soft tissues of the genitourinary system. The protocols CBCT-III and CBCT-IV were provided by the manufacturer. The protocol parameters were not changed for individual scans. The implemented automatic exposure control (AEC) modulates both tube current and tube voltage to keep detector entrance dose constant during scans. A standard focus-detector distance of 118 cm was chosen, resulting in an axial scan length of approximately 15 cm. The focus spot size was 0.2 mm. Conventional CT was performed at an MSCT with 16-detector rows (Emotion 16; Siemens Healthcare, Forchheim, Germany). Seven protocols were investigated (Table 2). Protocols MSCT-I and MSCT-VI to MSCT-VII are clinically used in abdominal low-dose and standarddose CT examinations of patients with suspected renal colic or urothelial cancer, respectively.16,17 Because MSCT protocols are usually adjusted to the specific needs of the institution and, thus, differ from site to site, many different protocols covering a large range of

pitch-corrected computed tomographic dose index values were used to study the relation between image quality and applied dose. Apart from protocol MSCT-VII, the pixel size was fixed to approximately 1 mm in the different protocols for a better comparison of image quality. Therefore, protocol MSCT-VII with a pixel size of 0.75 mm was used for dose measurements only. The AEC should provide images with a diagnostically adequate image quality considering the patient dose that was predetermined via a reference current-time-product Qref. It was turned on in 5 scan protocols. For the low-dose protocol MSCT-I and a scan of the International Commission on Radiological Protection (ICRP) reference person,18 the effective current-time-product Qeff was modulated to a value larger than the predetermined value Qref. For the other protocols, the current-time product was decreased by the AEC. For protocols MSCT-IV to MSCT-VII, a voltage of 130 kV, that is, the reference voltage of the Emotion 16-scanner, was used. The focus spot size of the MSCT scanner was 0.95 mm. A scan length of approximately 15 cm and a pitch of 0.8 were used for the measurements. Hounsfield unitsYcalibrated image matrices with 512 512 pixels were reconstructed.

Dose Measurements Dose measurements were performed with thermoluminiscent detectors (TLD; Bicron-Harshaw, Cleveland, OH). Each TLD was individually calibrated for a known absorbed dose in air using a conventional x-ray equipment with tube potentials and (equivalent) aluminum filters as used for the measurements at the CBCTand MSCT.

TABLE 2. Protocols Used at the MSCT System for Genitourinary Imaging No. Protocol MSCT-I MSCT-II MSCT-III MSCT-IV MSCT-V MSCT-VI MSCT-VII

Voltage, kV(p)

Qeff, mAs

Qref, mAs

CTDIvol, mGy

Kernel*

Pixel Size, mm†

Slice Thickness, mm‡

80 80 80 130 130 130 130

43 80 150 35 61 142 82

38 159 V 38 159 V 110

1.38 2.57 4.79 4.44 7.71 17.77 10.29

B30s/B50s/B70s B30s/B50s/B70s B30s/B50s/B70s B30s/B50s/B70s B30s/B50s/B70s B30s/B50s/B70s B30s

0.977 0.977 0.977 0.977 0.977 0.977 0.75

1.0 1.0 1.0 1.0 1.0 1.0 5.0

*The names of the convolution kernels stand for a soft (B30s), medium-hard (B50s), and hard (B70s) kernel. †Pixel size of the reconstructed image. ‡Slice thickness of the reconstructed image. CTDIvol indicates pitch-corrected computed tomographic dose index; Qeff, effective current-time-product; Qref, reference current-time-product when using the automatic exposure control.

676

www.investigativeradiology.com

* 2014 Lippincott Williams & Wilkins




The doses were measured for all available protocols of the CBCT system using approximately 200 TLDs distributed at the surface and in the interior of the anthropomorphic Alderson-Rando phantom (Alderson Research Laboratories Inc, Long Island City, NY) laid in supine position on the table of the imaging system. The phantom, which represents the ICRP reference man with a body weight of 73.5 kg and a trunk length of 68 cm, consists of a human skeleton embedded in different plastic materials that are radioequivalent to human soft tissues. The phantom is transected into 36 transaxial cross sections with a thickness of 2.5 cm each. Holes drilled in the cross sections on a 3 cm 3 cm grid were plugged by tissue-equivalent holder pins for TLDs. The used holes could uniquely be assigned to specific organs and tissues. The TLD positions were carefully selected in the region of interest (ROI) for genitourinary scans, considering an additional range potentially affected by scatter radiation. At least 2 TLDs were located in each organ. Equivalent organ doses of the brain, thyroid, and salivary glands, which received no direct exposure, were not measured. Equivalent organ doses were obtained from TLD measurements by taking the mean of the dose values recorded by the different TLDs within the specified organs. For largely extended organs (lung, skin, bone, and red bone marrow), equivalent doses were estimated using the scheme presented by Huda and Sandison.19 The TLDs at the surface of the phantom were used to sample the nonuniform local skin dose distribution. Effective dose values were determined from equivalent dose values using the tissue weighting factors of ICRP Publication 103.20 The total uncertainty of the dose measurement with the TLDs was estimated to be 6%, taking into account the statistical uncertainty of repeated TLD readings (3%) as well as systematic uncertainties arising from the energy dependence of the TLDs for the photon energies used for CT imaging (3%), the dependence of the TLD response on the direction of the incident radiation (3%), and the uncertainty in the calibration of an ionization chamber (2%) used for TLD calibration.21 The described TLD measurements are very laborious and time-consuming and, thus, were only carried out for 2 of the 7 MSCT protocols, that is, for MSCT-VI and MSCT-VII. Instead, the PC program CT-EXPO22 was used to compute equivalent organ doses for the MSCT protocols summarized in Table 2 from the respective scan range, beam collimation, pitch, as well as the effective pitch-corrected computed tomographic dose index and dose-length product listed in

Dose and Image Quality of CBCT and MSCT

the dose report of the scan. CT-EXPO is based on Monte Carlo data published by Zankl et al23 and has been described in detail elsewhere.24,25 It takes into account the overscanning effect and the longitudinal tube current modulation of many different scanner types, including the Emotion 16-scanner. The program was validated in a previous phantom study.26 According to this study, a random uncertainty of 11% was assumed for the results of CT-EXPO. To correct for potential systematic deviations between TLD measurements and dose computations with CT-EXPO, a calibration factor was determined from organ dose values obtained by both methods for MSCT protocols VI and VII. For this purpose, the calibration factor was determined from the mean ratio of measured and computed dose values of 8 organs within the beam (kidneys, adrenals, stomach, liver, gall bladder, spleen, small intestine, and colon).

Assessment of Image Quality Phantom Measurements For the assessment of image quality, image characteristics such as the contrast-to-noise ratio (CNR), the noise-power spectrum (NPS), and the modulation transfer function (MTF) were determined from cross-sectional CT images of a CT performance phantom (Fluke Corporation, Everett, WA). This so-called American Association of Physicists in Medicine (AAPM) phantom27 consists of a cylindrical, acrylic water container with a diameter of 21.6 cm containing different inserts to determine various image characteristics. The CT number insert, which was used to determine contrast and noise, consists of 5 cylinders with densities of 1.19, 1.05, 1.20, 0.95, and 1.10 g/cm3. The cylinders have a diameter of 0.45 cm each and are orientated parallel to the rotational axis of the phantom. The MTF insert consists of five 0.3-mmYthin tungsten wires. The wires were orientated parallel to the phantom axis at different distances from the axis to determine the dependence of the MTF from the measuring position within the image. Figure 1 shows CBCT images of both the CT number and MTF insert. The phantom was carefully positioned along the rotational axis of the CT by using the laser-light for patient positioning. Using a conventional x-ray device, it was found that the absorption of soft (80 kV[p], 5-mmYthick aluminium filter) and hard (120 kV[p], 8-mmYthick aluminium filter) radiation in the AAPM phantom is approximately identical to the absorption by the AldersonRando phantom in anterior-posterior projection. To ensure identical

FIGURE 1. Cone-beam CT images of the CT number insert (left) and MTF insert (right) of the AAPM phantom. Contrast-to-noise ratio was determined from ROIs located within the solid circles and the dashed circular arcs within and around the cylinders of the CT number insert. Noise-power spectra were derived from the quadratic ROIs around the cylinders. Quadratic ROIs in the CT image of the MTF insert were used to determine the MTF. Because of a small FOV, the outer edge of the phantom is truncated. * 2014 Lippincott Williams & Wilkins



677


Schegerer et al

absorption in both phantoms for all projection angles, which is a prerequisite for interrelating image quality and dose for different CT protocols, the AAPM phantom was placed between 2 shells of Perspex with variable thicknesses for different projection angles. The Perspex shells enclosed the AAPM phantom over the whole scan length of approximately 15 cm. Two shell pairs with different thicknesses (maximum of 3.6 and 6.9 cm, respectively) guaranteed the adaption to soft and hard radiation (Fig. 2). Contrast-to-noise ratio, NPS, and MTF were determined from the cross-sectional images. Because noise and spatial resolution also depend on pixel size and slice thickness of the reconstructed image matrix, which are smaller in CBCT as compared with MSCT because of different field of views (FOV), CNR, NPS, and MTF were additionally determined from rebinned CBCT images, that is, images resized by merging 2 pixels in each dimension using in-house postprocessing software. The pixel size of the ‘‘resized’’ CBCT images was almost identical with that of the MSCT images (0.98 mm 0.98 mm 0.98 mm vs 0.977 mm 0.977 mm 1.0 mm; Tables 1 and 2). A direct comparison of MSCT images with resized CBCT images is thus reasonable despite different FOVs. Image characteristics derived from these images are labeled ‘‘resized’’ images.

Contrast-to-Noise Ratio For the computation of CNR, circular ROIs with a radius of 14 pixels (7 pixels for the resized images) were placed over the cylinders of the CT number insert. To determine the background density, the cylinders were additionally circumscribed by 2 arcs with radii of 32 pixels and 35 pixels (16 and 18 pixels for the resized images). The arithmetic mean of pixel values mcyl within the ROIs placed in the cylinders (solid circle in Fig. 1) was subtracted from the arithmetic mean of pixel values mbg from the ROIs of the background (dashed arcs in Fig. 1). This difference was subsequently divided by the root of the quadratic sum of standard deviations R of the corresponding pixel values: mcyl jmbg CNR ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : R2cyl þ R2bg To reduce the statistical uncertainty, CNRs were measured for 32 parallel CT images of each protocol and convolution kernel, at least. Images showing artifacts were not considered. The relationship

FIGURE 2. The AAPM phantom and Perspex shells used to adapt to the radiation quality (hard radiation [left]; soft radiation [right]). 678



between (effective) dose and CNR for the same protocol was approximated by a quadratic function determined by a least-square fit.28

Noise-Power Spectrum The computation of the NPS is based on the Fourier transform of the deviation of pixel values from a mean signal within an ROI to determine the variance of noise power present at each spatial frequency. The magnitude of the NPS reflects the degree of randomness at each spatial frequency. The shape of the NPS reveals where the noise power is concentrated in frequency space: the maximum of the NPS at low frequencies results from coarse-grained noise, whereas the maximum at high-frequency noise power results from finer grain noise and a tangling background.29,30 By taking into account both the variance and spatial characteristics of the image noise, the NPS is a more thorough noise descriptor than pixel standard deviation. The 32 32 pixel-sized ROIs outside the cylinders of the CT number insert (Fig. 1) were used for computing the NPS. A 2-dimensional NPS was derived from each ROI using the standard approach.31Y33 To eliminate any systematic variation of image signal, a 2-dimensional polynomial function was first fitted to and subtracted from each ROI. The ROIs were potentially affected by beam hardening artifacts caused by the adjacent cylinders of the CT number insert. However, because the same procedure was used for all CBCTand MSCT images, the same putative artifacts appear in all NPS. Finally, 1-dimensional NPS curves were derived from the 2-dimensional NPS as described by Dobbins et al.34 As noise power proportionally decreases with increasing dose, the NPS determined for each protocol and convolution kernel were multiplied with the corresponding effective dose for normalization. This procedure allows a separate evaluation of the effects of the system, convolution kernel, and voltage on noise. To determine whether the NPS depends on the measuring position, NPS curves were separately computed for 16 ROIs of the central image region and for 20 ROIs of the outer image region (Fig. 1). The NPS of the 16 inner and that of the 20 outer ROIs were averaged for each protocol and convolution kernel to reduce statistical uncertainty. To study the potential dependency of NPS from image position within the broad cone beam, that is, along its rotational axis,35 the NPS were additionally determined for varying positions of the CT number insert within the cone beam in subsequent scans.

Modulation Transfer Function The spatial resolution of the CT images was measured from 12 12 pixel-sized ROIs around the five 0.05-mmYthin tungsten wires of the MTF insert (Fig. 1). The ROI data were spectrally analyzed using the Fourier transform as described in previous studies.36Y39 The resulting MTF provides a complete, quantitative description of (noisefree) system resolution and represents the capability of the system to show the contrast of objects of different finenesses.40,41 For each measurement, the MTF was determined for 35 parallel images, at least, and averaged to reduce statistical uncertainty. For CT images, which were reconstructed using the filtered back projection method, previous studies found that MTF decreases with increasing distance of the measurement position from the rotational center.42 To take this aspect into account, the MTFs were determined from wire profiles at the central and outer image regions (Fig. 1). In addition, the MTFs were determined at different axial positions along the central wire. For a convenient and clear comparison of the MTFs derived from the CBCT and MSCT images acquired with different CT protocols and at different image positions, the arithmetic mean Gf50&109 of spatial frequencies at the 50% and 10% MTF values were determined (Fig. 6 for further explanation). These values were found by linear interpolation of adjacent spatial frequencies. This approach was used, for example, by the Medical Devices Agency group on Imaging Performance Assessment of CT Scanners (ImPACT).43 * 2014 Lippincott Williams & Wilkins




RESULTS The surface dose distributions reach a maximum dose of approximately 3 mGy, 15 mGy, 76 mGy, and 190 mGy for the protocols CBCT-I to CBCT-IV. Although the surface doses were almost constant at dorsal projections, they strongly decreased to a few milligrays at ventral projections because of the partial rotation of CBCT by approximately 200 degrees around the back of the patient. The increase in the local skin dose of protocols CBCT-II to CBCT-IV with respect to the skin dose of protocol CBCT-I corresponds to an increase in the dose area products (right column of Table 1). At the MSCT system, the local skin dose at the surface of the cross sections is almost constant. The local skin dose at dorsal positions of the midsagittal plane of the phantom is shown in Figure 3. The scan with an axial length of approximately 15 cm does not only compass the kidneys but also covers adjacent radiosensitive organs such as the stomach and liver. Table 3 lists organ dose values measured at the Alderson-Rando phantom for the protocols CBCT-I to CBCT-IV and MSCT-VI to MSCT-VII. The mean ratio of measured and computed equivalent dose values of organs within the beam is 1.06 T 0.09 (mean T SEM). The deviations of computed from measured dose values result from the different internal configuration and shape of organs of the numerical and the anthropomorphic phantom.22 To correct for this systematic deviation, organ doses computed for protocols MSCT-I to MSCT-V were multiplied with the calibration factor. The effective doses of these protocols were 0.74 mSv, 1.43 mSv, 1.85 mSv, 1.70 mSv, and 3.02 mSv, respectively.

Assessment of Image Quality Contrast-to-Noise Ratio The effective doses are plotted in Figure 4 versus the CNR values averaged for the 5 cylinders of the CT number insert for different CBCT and MSCT protocols and convolution kernels. Contrastto-noise ratio strongly depends on the density and location of the


cylinders in the FOV resulting in the relatively large standard deviation in CNR shown in Figure 4. Data of CBCT deviate substantially from the assumed quadratic curves because of the erroneous assumption that only the tube current was changed, although the tube voltage that can affect contrast, too, were additionally changed during the CBCT scans. The CNR values derived from the nonYresized CBCT images were in the range determined for MSCT images reconstructed with the hard and medium-hard convolution kernel. For a given CNR, the lowest effective dose was obtained for the MSCT scans reconstructed with the soft convolution kernel. The largest effective dose values occur for the MSCT scans reconstructed with the hard convolution kernel. The effective dose that has to be applied to get the same CNR is lower by a factor of approximately 5 for the MSCT images reconstructed with the soft convolution kernel, compared with the CBCT measurement.

Noise-Power Spectrum The effect of pixel/image resizing, convolution kernel, tube voltage, image region, and the imaging system on image noise is shown in Figure 5, where the dose-normalized noise power spectra are plotted. As expected, the NPS values derived from the resized CBCT images decreased compared with the corresponding NPS values of the nonYresized images. The effect of the convolution kernel of the MSCT scans is apparent at large spatial frequencies where the hard convolution kernel increases noise. Relatively constant NPS curves, corresponding to white noise, were obtained for the MSCT images reconstructed with the medium-hard convolution kernel and the CBCT images. The soft convolution kernel suppresses image noise, particularly at large spatial frequencies. Although the tube voltage of MSCT scans hardly affects the NPS (effects within the plotted error bars), the NPS of CBCT scans at tube voltages of 103 kV(p) and 115 kV(p) (protocols CBCT-I and CBCT-VI) are more than a factor of 2 larger than the NPS derived at tube voltages of 85 of 90 kV(p).

FIGURE 3. Skin dose at different dorsal positions of the anthropomorphic phantom plotted versus the number of the cross sections (thickness of 2.5 cm) of the phantom. Black dotted, solid, dashed, and dotted-dashed curves represent the measurements for protocols CBCT-I to CBCT-IV, respectively. The thicker gray curve provides the skin dose measured for protocol MSCT-VII. The position of the lung and the kidneys of the Alderson-Rando phantom is indicated by the light- and dark-grayYshaded regions, respectively. * 2014 Lippincott Williams & Wilkins



679


Schegerer et al


TABLE 3. Equivalent Organ (in millisieverts) and Effective Doses (in millisieverts) Obtained for Protocols CBCT-I, CBCT-IV, MSCT-VI, and MSCT-VII MSCT-VI Organ Brain Thyroid Esophagus Thymus Breast Lung Stomach Liver Spleen Kidneys Adrenals Small intestine Bladder Colon Uterus Gonads Skin ABM Effective dose

MSCT-VII

CBCT-I

CBCT-II

CBCT-III

CBCT-IV

Measured

Computed

Measurement

Computed

0.00 0.00 0.07 0.00 0.02 0.08 0.99 0.89 0.76 2.29 0.88 0.59 0.08 0.59 0.07 0.04 0.20 0.28 0.35

0.00 0.00 0.44 0.00 0.09 0.34 4.24 3.25 3.63 9.95 3.69 2.46 0.40 2.46 0.44 0.23 0.89 1.29 1.48

0.00 0.00 2.00 0.00 0.49 1.41 20.17 16.79 18.32 46.36 19.57 10.39 1.35 10.39 1.47 0.82 4.60 5.35 6.74

0.00 0.00 4.89 0.00 0.90 4.17 48.55 42.84 49.08 117.33 56.84 26.34 4.03 26.34 4.97 2.49 10.46 14.22 17.10

0.00 0.00 3.69 0.00 1.09 2.24 26.58 22.61 16.15 27.01 17.72 13.83 2.60 13.83 2.44 1.27 2.80 2.20 8.01

V V V V V V 22.35 21.70 22.30 23.05 17.90 13.85 V 9.92 V V V V 7.70

0.00 0.00 1.26 0.00 0.37 0.74 11.24 8.78 7.57 11.50 8.63 6.33 1.32 6.33 1.25 0.66 1.13 2.20 3.56

V V V V V V 12.10 11.40 9.15 12.85 9.15 5.70 V 4.45 V V V V 3.65

For protocols MSCT-VI and MSCT-VII as well as organs within the beam, both measured and computed doses were listed. The tissue weighting factors of ICRP Publication 103 were used to determine effective doses.16 ABM indicates active bone marrow; CBCT, cone-beam computed tomography; ICRP, international commission on radiological protection; MSCT, multislice spiral computed tomography.

FIGURE 4. Effective doses plotted versus the CNR. Diamonds, triangles, and squares represent measurements at the MSCT system using the soft, medium-hard, and hard convolution kernel, respectively. Black unfilled/filled circles stand for CNR measurements using the original/resized CBCT images. The cross in the lower right corner represents the average statistical uncertainty of the data. The relationship between (effective) dose and CNR for the same protocol was approximated by a quadratic function. 680







FIGURE 5. A to D, Normalized NPS derived from (resized) CT images of the protocols CBCT-I to CBCT-IV and MSCT-I to MSCT-VI (Tables 1 and 2). The NPS of the nonYresized CBCT images are plotted in the A and B, whereas the NPS of the MSCT images and resized CBCT images are shown in the C and D lower figures. The NPS were derived from central (left) and outer image regions (right; Fig. 1). The vertical error bars represent standard deviations of the normalized NPS derived from measurements at the MSCT where the same convolution kernel but different tube voltages were applied. Dotted, solid, dashed, and dotted-dashed black lines represent the NPS of protocols CBCT-I, CBCT-II, CBCT-III, and CBCT-VI, respectively. Diamonds, triangles, and squares represent measurements at the MSCT system using the soft, medium-hard, and hard convolution kernel, respectively. Black unfilled/filled circles stand for CNR measurements using the original/resized CBCT images.

The NPS determined at the outer regions of the phantom were only half as high as those at the inner regions, particularly at large spatial frequencies. A significant dependency of NPS on the position of the cross-sectional image within the 15-cmYlong phantom could not be found (not shown in Fig. 5). The NPS of images derived by protocols CBCT-I and CBCT-IV at tube voltages of 115 kV(p) and 103 kV(p), respectively, correspond to the NPS of MSCT images reconstructed with the medium-hard kernel, whereas the NPS of protocols CBCT-II and CBCT-III, using voltages of 85/90 kV9(p), are in the range of the NPS of MSCT images reconstructed with the soft convolution kernel.

Modulation Transfer Function For the investigated protocols, convolution kernels, and image regions, the results are shown in Figure 6. In contrast to CNR and NPS, spatial resolution is independent of dose and image noise but depends on the convolution kernel. However, noise determines the uncertainty of MTF measurements. Therefore, the uncertainty of MTF is large for the MSCT * 2014 Lippincott Williams & Wilkins

images reconstructed with the hard convolution kernel. According to the results summarized in Table 4, Gf50&10 9 values are lower for softer convolution kernels and for nonYcentral image regions. A clear dependency of the MTF of the CBCT system from the image position within the axial scan length of 15 cm, that is, parallel to the rotational axis of the phantom, could not be found. Although the intrinsic pixel size and slice thickness of CBCT images (without resizing) are half of those of MSCT images, the MTF of CBCT is lower than the MTF derived from the MSCT images reconstructed with the hard convolution kernel. For the considered CBCT system, the Gf50&10 9 values derived from the resized images are comparable with those derived from MSCT images reconstructed with the medium-hard convolution kernel.

DISCUSSION Cone-beam CT may develop to become a standard technique for many interventional and operative procedures, for example, in urolithiasis and percutaneous nephrolithotomy, and has already replaced, to some extent, conventional MSCT scans carried out in the radiology department.44 The aim of this study was thus to determine www.investigativeradiology.com


681


Schegerer et al

FIGURE 6. Mean MTF derived from CBCT (solid black lines with unfilled/filled circles) and MSCT images (dotted, dashed, and dotted-dashed gray curves with diamonds, triangles, and squares, respectively) of the profiles of the central 0.05-mmYthin tungsten wire of the MTF insert. Diamonds, triangles, and squares represent measurements at the MSCT system using the soft, medium-hard, and hard convolution kernel, respectively. Black unfilled/filled circles stand for CNR measurements using the original/resized CBCT images. The MTF values at a spatial frequency of f = 0 were omitted because they are strongly affected by noise. The spatial frequencies f50 and f10 correlate to the MTF values of 0.5 and 0.1, respectively. In the plot, the f50 and f10 values of the nonYresized CBCT images are shown.

in detail radiation dose and quality of images produced by a CBCT system used for abdominal and genitourinary imaging. The results were compared with the dose and image quality determined using an MSCT scanner. For the assessment of image quality, the physical parameters NPS, MTF, and CNR were used to quantify how noisy/ sharp the images appear to the viewer. This is an established approach that has been used in previous studies for the comparison of imaging devices with different detector systems.45Y47 As compared with conventional radiography and fluoroscopyguided abdominal and genitourinary examinations, CBCT exposes patients to higher radiation doses.48Y52 The relatively high equivalent organ doses for CBCT scans are confirmed by the recent study of


Kwok et al,53 who also measured equivalent organ doses in CBCT for different abdominal protocols. Therefore, it is indispensable to carefully justify and optimize the CBCT protocols predefined by the manufacturer to the specific clinical needs of different types of examinations. The protocols CBCT-I and CBCT-II investigated in this study were developed to minimize dose and, simultaneously, fulfill the requirements on image quality for specific urogenital examinations. The parameter setting of CBCT-I, for instance, was optimized to search for, and assess, renal concrements, that is, hard-contrast objects and percutaneous nephrolithotomy punctures. The image quality resulting from the relatively low effective exposure is sufficient for these medical issues. The effective dose of this low-dose protocol corresponds to the dose applied for conventional radiographs of the pelvis. The parameter settings of the 4 CBCT protocols are not further adapted to individual patients in clinical routine. Contrarily, MSCT scanners offer much more protocols adapted to different medical issues. To further optimize CBCT scans with respect to patient exposure, more protocols have to be developed and offered in CBCT systems. To further lower exposures and improve image quality, experienced physicians (in cooperation with a medical physicist) should individually adapt the acquisition and image reconstruction parameters of CBCT systems (including the convolution kernel) to the particular clinical issue and the patient’s size. In addition to protocol optimization, further technical developments, such as detectors with better soft contrast detectability, that is, lower image noise in the highYkilovolt (peak) range, could further lower dose (Fig. 5). Iterative image reconstruction algorithms that have not been established in CBCT systems, so far, have already shown their dose-saving potential with MSCT systems.54 In any case, however, the collimation of the radiation field has to be adapted to the clinically relevant body region to reduce patient exposure.4 In case of genitourinary CBCT examinations, for instance, the dose could be more than half when solely the kidneys were to be examined (Fig. 3). In contrast to MSCT, where data are acquired in helical mode, the investigated CBCT system acquires image data from 1 partial rotation around the patient of approximately 200 degrees. Therefore, (entrance) dose is increased relative to effective dose in CBCT scans to obtain a similar image quality (eg, CNR) as with MSCT. In fact, the ratio of the average local skin dose and effective dose of protocol CBCT-II (approximately 15 mGy/1.57 mSv) is approximately 3 times larger than the corresponding ratio of protocol MSCT-VII (10.6 mGy/3.6 mSv). In case of complicated procedures where patients are scanned several times, the skin may locally be exposed to relatively high doses of many hundreds of millisievert (Fig. 3). The transaxial FOV of CBCT was relatively small. As a consequence, the image quality of the reconstructed CBCT images may be hampered by incorrectly reconstructed density values and truncation artifacts at border regions of the FOV (Fig. 1).15 According to our results, however, these artifacts do not affect the central image

TABLE 4. Gf50&109 Values (SD) Derived From MTFs Determined for Different Protocols at the CBCT and MSCT Scanner Gf50&109 Protocol/Region CBCT-I and CBCT-IV, nonresized CBCT-I and CBCT-IV, resized Convolution kernel MSCT-I and MSCT-VI

Hard 0.69 (0.81)

Inner Image Region

Outer Image Region

0.73 (0.11) 0.58 (0.15) Medium-hard 0.56 (0.22)

0.68 (0.13) 0.55 (0.19) Medium-hard 0.55 (0.17)

Soft 0.48 (0.07)

Hard 0.72 (0.72)

soft 0.45 (0.09)

CBCT indicates cone-beam computed tomography; MSCT, multislice spiral computed tomography; MTF, modulation transfer function.

682






region. For examinations of body regions close to the surface, the isocenter can be moved. The simultaneous analysis of dose and image quality makes it possible to compare diverse diagnostic x-ray systems, such as the MSCT and CBCT. To obtain CBCT images with pixel sizes and a slice thickness similar to the ones of MSCT images and, thus, to allow a direct comparison of noise power and spatial resolution, CBCT images were resized in the present study by merging 2 pixels in each dimension, resulting in a CNR increase and noise decrease. Such an approach can be used when imaging low-contrast lesions where high spatial resolution is not required. In that case, the spatial resolution is similar to that reached in MSCT using a medium-hard convolution kernel, but the CNR and NPS are as good as the corresponding quantities derived from MSCT images reconstructed with the soft convolution kernel. The relatively flat NPS of CBCT images guarantees a homogenous graininess of the images. Comparing the NPS obtained for different tube voltages, the noise power of images acquired at tube voltages of 85/90 kV(p) is approximately half of that of images measured at 115 kV(p) and even at 103 kV(p). This result confirms a previous finding by Fahrig et al14 that the contrast detectability of CBCT systems reaches a maximum at lower tube voltages compared with MSCT. Cone-beam CT systems should thus preferably operate at approximately 90 kV(p), in particular when low-contrast lesions are imaged. For tube voltages lower than approximately 90 kV(p), in turn, noise power increases resulting from a degradation of detector performance. Furthermore, softer (or lower energy) radiation at approximately 60 kV(p) is absorbed more strongly; thus, noise increases relative to images at higher kilovolt (peaks).55 Softer radiation quality also increases susceptibility to image artifacts because of beam hardening or photon starvation, for example, when using metallic pistons or high concentrations of contrast agents. Therefore, soft radiation quality is not used in the low-dose protocol CBCT-I used for percutaneous nephrolithotomy punctures. In contrast, a clear dependency of image noise on tube voltage could not be found in case of the MSCT images: the NPS derived for the tube voltages of 80 kV(p) and 130 kV(p) were identical within the error bars (Fig. 5). The contrast detectability of detectors used in MSCT scanners does not depend from voltage as strongly as the detectors of CBCT systems do. Previous studies have found that CT systems with large flat panel detectors and, thus, large cone beams are affected by a large portion of noise from scattered radiation, particularly in the central part of the cone beam.35 Furthermore, x-rays of large cone beams transmit different distances through the patient, resulting in different absorption and so-called cone-beam artifacts. These artifacts, which are more pronounced for outer detector rows than for the central ones, get worse with an increasing cone angle.56 In this study, worsening of image quality, as characterized by the noise power and spatial resolution, could not be found in the CBCT images of the inserts of the AAPM phantom. Contrarily, because of the filtered back projection algorithm, there are lower NPS and MTF curves in the outer image regions relative to the central region. However, this degradation of image quality is within the error bars and, thus, does probably not affect diagnosis. The inverse of the spatial frequency Gf50&109j1 characterizes the minimum size of objects that can be resolved with the system. Using the hard, medium-hard, and soft convolution kernel, objects with sizes of Gf50&109j1 approximately equal to 1.4 mm, 1.8 mm, and 2.1 mm can theoretically be resolved with MSCT, respectively. The CBCT system theoretically resolves objects with sizes of 1.4 mm (1.7 mm for resized images). The critical size of urology structures to be detected is approximately 5 mm, and this can be resolved with both MSCT and CBCT scanners.57 The relatively low resolving capacity of CBCT in contrast to MSCT detectors results from blur caused by the missing reflective coating of detector elements and * 2014 Lippincott Williams & Wilkins


missing antiYscatter grid that minimize cross-talk and scatter radiation.58 Blur reduces spatial resolution and, at the same time, noise. Therefore, the MTF derived from CBCT is less noisy than that of the MSCT images reconstructed with the hard convolution kernel (Fig. 6), resulting in a potentially better detectability of (small and hard contrast) objects. The CNR and NPS of (nonresized) CBCT images were found to be superior to, or at least as good as, MSCT images reconstructed with the hard or even the medium-hard convolution kernel. The spatial resolution achieved in CBCT images is at least as good as that of MSCT images reconstructed with the medium-hard convolution kernel. A limitation of our study is the use of a motionless physical phantom to determine image characteristics. In clinical use, the relatively long rotation time of CBCT systems may result in a large susceptibility to organ and patient movements, that is, movement artifacts and, thus, in image degradation, particularly of spatial and low contrast resolution.7,59,60 REFERENCES 1. Missler U, Hundt C, Wiesmann M, et al. Three dimensional reconstructed rotational digital subtraction angiography in planning treatment of intracranial aneurysms. Eur Radiol. 2000;10:564Y568. 2. Hochmuth A, Spetzger U, Schumacher M. Comparison of three-dimensional rotational angiography with digital subtraction angiography in the assessment of ruptured cerebral aneurysms. AJNR Am J Neuroradiol. 2002;23:1199Y1205. 3. Song JK, Niimi Y, Brisman L, et al. Simultaneous bilateral internal carotid artery 3-D rotational angiography. Technical note. AJNR Am J Neuroradiol. 2004;25:1787Y1789. 4. Struffert T, Hertel V, Kyriakou Y. Imaging of cochlear implant electrode array with flat-detector CT and conventional multislice CT: comparison of image quality and radiation dose. Acta Otolaryngol. 2010;130:443Y452. 5. Rieber G, Rohkohl C, Lauritsch G, et al. Kardiale Anwendung der C-ArmComputertomographie. Radiologe. 2009;49:862Y867. 6. Moesler J, Dittrich S, Rompel O, et al. Flachdetektor-Computertomografie in der diagnostischen und interventionellen Kinderkardiologie. Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr. 2013;185:446Y453. 7. Meyer BC, Frericks BB, Albrecht TH, et al. Contrast-enhanced abdominal angiographic CT for intra-abdominal tumor embolization: a new tool for vessel and soft tissue visualization. Cardiovasc Intervent Radiol. 2007;30:743Y749. 8. Hirota S, Nakao N, Yamamoto S, et al. Conebeam CT with flat-panel-detector digital angiography system: early experience in abdominal interventional procedures. Cardiovasc Intervent Radiol. 2006;29:1034Y1038. 9. Guggenberger R, Winklhofer S, Spiczak JV, et al. In vitro high-resolution flatpanel computed tomographic arthrography for artificial cartilage defect detection: comparison with multidetector computed tomography. Invest Radiol. 2013;48:614Y621. 10. Guggenberger R, Fischer MA, Hodler J, et al. Flat-panel CT arthrography: feasibility study and comparison to multidetector CT arthrography. Invest Radiol. 2012;47:312Y318. 11. So¨derman M, Babic D, Homan R, et al. 3D roadmap in neuroangiography: technique and clinical interest. Neuroradiology. 2005;47:735Y740. 12. Richter G, Engelhorn T, Struffert T, et al. Flat panel detector angiographic CT for stent-assisted coil embolization of broad-based cerebral aneurysms. AJNR Am J Neuroradiol. 2007;28:1902Y1908. 13. Michel MS, Trojan L, Rassweiler JJ. Complications in percutaneous nephrolithotomy. Eur Urol. 2007;51:899Y906. 14. Fahrig R, Dixon R, Payne T, et al. Dose and image quality for a cone-beam Carm CT system. Med Phys. 2006;33:4541Y4550. 15. Mori S, Endo M, Komatsu S, et al. A combination-weighted Feldkamp-based reconstruction algorithm for cone-beam CT. Phys Med Biol. 2006;51:3953Y3965. 16. van der Molen AJ, Cowan NC, Mueller-Lisse UG, et al. CT urography: definition, indications and techniques. A guideline for clinical practice. Eur Radiol. 2008;18:4Y17. 17. Poletti PA, Platon A, Rutschmann OT, et al. Low-dose versus standard-dose CT protocol in patients with clinically suspected renal colic. AJR Am J Roentgenol. 2007;188:927Y933. 18. International Commission on Radiological Protection (ICRP). Adult Reference Computational Phantoms. Valentin J, ed. 2009; ICRP Publication 110. Ann. ICRP 39 (2). 19. Huda W, Sandison GA. Estimation of mean organ doses in diagnostic radiology from Rando phantom measurements. Health Phys. 1984;47:463Y467. 20. International Commission on Radiological Protection (ICRP). The 2007 Recommendations of the International Commmission on Radiological Protection. Valentin J, ed. 2007; ICRP Publication 110. Ann. ICRP 39 (2).



683


Schegerer et al

21. Lechel U, Becker C, Langenfeld-Ja¨ger G, et al. Dose reduction by automatic exposure control in multi-slice computed tomographyVcomparison between measurement and calculation. Eur Radiol. 2009;19:1027Y1034. 22. Stamm G, Nagel HD. CT-EXPO [computer program]. Version 2.2. Hannover/ Buchholz, Germany. 2013. 23. Zankl M, Panzer W, Drexler G. The calculation of dose from external photon exposures using reference human phantoms and Monte Carlo methods. Part VI: organ doses from computed tomographic examinations. GSF report. 1991;30:91. 24. Galanski M, Nagel HD, Stamm G. CT-Expositionspraxis in der Bundesrepublik Deutschland Ergebnisse einer bundesweiten Umfrage im Jahre 1999. Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr. 2000;173:1Y66. 25. Nagel HD, Galanski M, Hidajat N, et al. Radiation Exposure in Computed Tomography Fundamentals, Influencing Parameters, Dose Assessment, Optimization, Scanner Data, Terminology. 4th ed. Hamburg, Germany: CTB; 2002. 26. Brix G, Lechel U, Veit R, et al. Assessment of a theoretical formalism for dose estimation in CT: an anthropomorphic phantom study. Eur Radiol. 2004;14: 1275Y1284. 27. Fisher JR, Lin PJP, Butler P, et al. AAPM Report No. 60: Instrumentation requirements of diagnostic radiological physicists. Report of Task Group 4, Diagnostic X-Ray Imaging Committee. 28. Dobbins JT. Image quality metrics for digital systems. In: Beutel J, Kundel HL, van Metter RL, eds. Medical Imaging. Society of Photo-Optical Instrumentation Engineers (SPIE). Bellingham, Washington: Society of Photo-Optical Instrumentation Engineers (SPIE); 2000;1:161Y222. 29. Hanson KM. Detectability in computed tomographic images. Med Phys. 1979; 6:441Y451. 30. Wagner RF, Brown DG, Pastel MS. Application of information theory to the assessment of computed tomography. Med Phys. 1979;6:83Y94. 31. International Electrotechnical Commission (IEC). Medical electrical equipmentV characteristics of digital X-ray imaging devicesVPart 1: Determination of the detective quantum efficiency. 2003:IEC 62220-1. 32. Kijewski MF, Judy PF. The noise power spectrum of CT images. Phys Med Biol. 1987;32:565Y575. 33. Boone JM. X-ray production, interaction, and detection in diagnostic imaging. In: Beutel J, Kundel HL, van Metter RL, eds. Medical Imaging. Society of Photo-Optical Instrumentation Engineers (SPIE). Bellingham, Washington: Society of Photo-Optical Instrumentation Engineers (SPIE); 2000;1:1Y77. 34. Dobbins JT, Ergun DL, Rutz L, et al. DQE(f ) of four generations of computed radiography acquisition devices. Med Phys. 1995;22:1581Y1593. 35. Schlattl H, Tischenko O, Hoeschen C. Modeling of realistic raw data for image reconstruction: quantifying scattering noise in different CT geometries. In: Flynn MJ, Hsieh J, eds. Medical Imaging. Society of Photo-Optical Instrumentation Engineers (SPIE). Bellingham, Washington: Society of PhotoOptical Instrumentation Engineers (SPIE); 2006:614256Y614257. 36. Judy PF. The line spread function and modulation transfer function of a computed tomographic scanner. Med Phys. 1976;3:233Y236. 37. Logan KW, Hickey KA, Bull SR. Gamma camera MTFs from edge response function measurements. Med Phys. 1983;10:361Y364. 38. Blendl C, Fiebich M, Voigt JM, et al. Investigation on the 3 D geometric accuracy and on the image quality (MTF, SNR and NPS) of volume tomography

684


39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54.

55. 56. 57. 58. 59. 60.


units (CT, CBCT and DVT). Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr. 2012;184:24Y31. Le Heron JC, Poletti JL. Measurement of x-ray image intensifier sharpness in the x-ray department. Phys Med Biol. 1988;33:93Y104. Bischof CJ, Ehrhardt JC. Modulation transfer function of the EMI CT head scanner. Med Phys. 1977;4:163Y167. Rossmann K. Point spread-function, line spread-function, and modulation transfer function. Radiology. 1969;93:257Y272. Kwan AL, Boone JM, Yang K, et al. Evaluation of the spatial resolution characteristics of a cone-beam breast CT scanner. Med Phys. 2007;34:275Y281. Lewis M, Keat N, Edyvean S. 32 to 64 slice CT scanner comparison report version 14 [impactscan Web site]. 2006. Available at: http://www.impactscan.org. Accessed January 27, 2014. Tu¨rk C, Knoll T, Petrik A, et al. Guidelines on Urolithiasis. European Association of Urology. 2013. Samei E, Flynn MJ. An experimented comparison of detector performance for direct and indirect digital radiography systems. Med Phys. 2003;30:608Y622. Monnin P, Marshall NW, Bosmans H, et al. Image quality assessment in digital mammography: part II. NPWE as a validated alternative for contrast detail analysis. Phys Med Biol. 2011;56:4221Y4238. Marshall NW, Monnin P, Bosmans H, et al. Image quality assessment in digital mammography: part I. Technical characterization of the systems. Phys Med Biol. 2011;56:4201Y4220. Samara ET, Aroua A., Bochud FO. Swiss population exposure to radiation by interventional radiology in 2008. Health Phys. 2012;103:317Y321. Lipkin ME, Mancini JG, Zilberman DE, et al. Reduced radiation exposure with the use of an air retrograde pyelogram during fluoroscopic access for percutaneous nephrolithomy. J Endourol. 2011;25:563Y567. John BS, Patel U, Anson K. What radiation exposure can a patient expect during a single stone episode? J Endourol. 2008;22:419Y422. Nawfel RD, Judy PF, Schleipman AR, et al. Patient radiation dose at CT urography and conventional urography. Radiology. 2004;232:126Y132. Mettler FA, Huda W, Yoshizumi TT, et al. Effective doses in radiology and diagnostic nuclear medicine. Radiology. 2008;248:254Y263. Kwok YM, Irani FG, Tay KH. Effective dose estimates for cone beam computed tomography in interventional radiology. Eur Radiol. 2013;23:3197Y3204. Baker ME, Dong F, Primak A, et al. Contrast-to-noise ratio and low-contrast object resolution on full- and low-dose MDCT: Safire versus filtered back projection in a low-contrast object phantom and in the liver. AJR Am J Roentgenol. 2012;199:8Y18. Yang K, Kwan ALC, Huang SY, et al. Noise power properties of a cone-beam CT system for breast cancer detection. Med Phys. 2008;35:5317Y5327. Barrett JF, Keat N. Artifacts in CT: recognition and avoidance. Radiographics. 2004;24:1679Y1691. Parma MS. Kidney stones. BMJ. 2004;328:1420Y1424. KachelrieQ M. Fundamentals of cone-beam imaging. In: European Society of Radiologists. Insights into Imaging. 2013;4:78. Akpek S, Brunner T, Benndorf G., et al. Diagn Interv Radiol. 2005;11:10Y13. Orth RC, Wallace MJ, Kuo MD. C-arm cone-beam CT: general principles and technical considerations for use in interventional radiology. J Vasc Interv Radiol. 2008;19:814Y820.



Radiation dose and image quality with abdominal computed tomography with automated dose-optimized tube voltage selection.

A comparative study for image quality and radiation dose of a cone beam computed tomography scanner and a multislice computed tomography scanner for paranasal sinus imaging.

Stereolithographic biomodelling of pulmonary hilum by multislice computed tomography imaging.

Artifacts in orthodontic bracket systems in cone-beam computed tomography and multislice computed tomography.

The image quality of ion computed tomography at clinical imaging dose levels.

Daily patient setup error in prostate image guided radiation therapy with fiducial-based kilovoltage onboard imaging and conebeam computed tomography.

Abdominal perfusion computed tomography.

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?

Computed tomography in abdominal trauma: an audit of usage and image quality.

Evaluation of abdominal abscess with computed tomography.

Radiation dose and image quality in K-edge subtraction computed tomography of lung in vivo.

Image quality and dose assessment in inner ear computed tomography imaging with a flat panel-based system.

Regional survey of image quality and radiation dose in computed tomography examinations in Saudi Arabia.

Image Quality of 256-Slice Computed Tomography for Coronary Angiography.

Computed Tomography Number Measurement Consistency Under Different Beam Hardening Conditions: Comparison Between Dual-Energy Spectral Computed Tomography and Conventional Computed Tomography Imaging in Phantom Experiment.

computed tomography and conventional imaging with computed tomography scans.

Picture quality of computed tomography.

Theory of image reconstruction in computed tomography.

Comparing the image quality of a mobile flat-panel computed tomography and a multidetector computed tomography: a phantom study.

Collimation and Image Quality of C-Arm Computed Tomography: Potential of Radiation Dose Reduction While Maintaining Equal Image Quality.

Assessment of bioabsorbable scaffolds by multislice computed tomography angiography.

Evaluation of myocardial fat deposition by multislice computed tomography.

Imaging dose from cone beam computed tomography in radiation therapy.

Measurement of skin dose from cone-beam computed tomography imaging.