Medical Engineering & Physics 35 (2013) 18371842

Contents lists available at ScienceDirect

Medical Engineering & Physics

journal homepage: www.elsevier.com/locate/medengphy

Technical note

Dynamic imaging with dual-source gated Computed Tomography (CT): Implications of motion parameters on image quality for wrist imaging
Puay Yong Neo a , Ita Suzana Mat Jais a , Christoph Panknin c , Chin Cheung Lau a , Lai Peng Chan d , Kai Nan An e , Shian Chao Tay a,b,
a

Wrist Analysis Research Laboratory, Singapore General Hospital, Singapore Department of Hand Surgery, Singapore General Hospital, Singapore c Siemens Healthcare, Forchheim, Germany d Department of Diagnostic Radiology, Singapore General Hospital, Singapore e Biomechanics Laboratory, Mayo Clinic College of Medicine, Rochester, MN, USA
b

a r t i c l e

i n f o

a b s t r a c t
Objective: Dynamic Computed Tomography (CT) promises insights into the pathophysiology of carpal instability by recording images of the carpus while it is in motion. The purpose of this study was to investigate the effect of motion velocity on image quality for dynamic carpal imaging applications using a clinical dual-source CT (DSCT) scanner. Methods: A phantom with targets in the axial, coronal and sagittal planes was attached to a motion simulator and imaged using a 64-slice DSCT scanner. Data was acquired when the phantom was stationary and during periodic linear motion. Spatial resolution, motion artifacts and banding artifacts were assessed. Results: Mean spatial resolution was 0.82 mm at 36 mm/s and 0.79 mm at 18 mm/s. Banding artifacts were mild at 36 mm/s and minimal at 18 mm/s. Motion artifacts were minimal at motion velocity of up to 36 mm/s in both the coronal and sagittal planes. Axial plane motion artifacts were moderate at 36 mm/s and mild at 18 mm/s. Discussion: Sub-millimeter resolution is achievable with commercially available DSCT scanners with mild to moderate amounts of motion artifacts at velocities of 18 mm/s and 36 mm/s respectively. 2013 IPEM. Published by Elsevier Ltd. All rights reserved.

Article history: Received 8 March 2012 Received in revised form 17 April 2013 Accepted 18 May 2013 Keywords: 4DCT Real-time imaging DSCT Dual source CT Computed tomography Image quality

1. Introduction ECG gated computed tomography (CT) imaging has become a valuable clinical tool for assessing coronary artery disease. In the same way, gated CT imaging can be a valuable tool for dynamic musculoskeletal imaging of small joints such as the wrist. Dynamic musculoskeletal CT imaging was referred to as 4-dimensional (4D) imaging (i.e. 3D + time) in our previous study performed using a single-source CT (SSCT) [1]. The ability to image 3D joint structures during real time motion has always been desirable as it extends the diagnostic capabilities of imaging modalities to include motion abnormalities. Furthermore, in 3D imaging, we can perform a comprehensive evaluation of the wrist motion in 3

Corresponding author at: Department of Hand Surgery, Singapore General Hospital, Outram Road, Singapore 169608, Singapore. Tel.: +65 63264588; fax: +65 62273573; mobile: +65 92303460. E-mail addresses: tay.shian.chao@sgh.com.sg, tay sc77@yahoo.com (S.C. Tay).

directions (exion-extension, radioulnar deviation, and pronationsupination). Wrist instabilities can be categorized as dynamic and static. Injuries to the scapholunate interosseous ligament (SLIL), are the most common cause of dynamic wrist instabilities. Patients with such injuries, despite experiencing wrist pain during motion, do not demonstrate any abnormalities on conventional (static) radiographic examinations. With 4D imaging, clinicians may be able to diagnose these conditions more effectively. For a complex joint such as the wrist which exhibits 6 degrees of freedom in motion, dynamic wrist CT imaging has far-ranging implications. In recent years, studies carried out by Carelson B et al. and Founami M et al. [2,3], have demonstrated the feasibility in performing in vivo dynamic studies of the carpus. In both studies, the authors performed 4D imaging of the wrists using rotational X-ray (4D-RX) imaging system. However, this approach has its drawbacks. In 4D-RX imaging, the joint has to be moved through many cycles of motion, to acquire 4D images, as compared to using CT. In addition to a better understanding of wrist kinematics, the long-term objective of this study is to address the

1350-4533/$ see front matter 2013 IPEM. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.medengphy.2013.05.009

1838

P.Y. Neo et al. / Medical Engineering & Physics 35 (2013) 18371842

clinical problem of diagnosing dynamic carpal instabilities. Current methods for imaging carpal motion are limited to either twodimensional (2D) video uoroscopy, or step-and-shoot series of static three-dimensional (3D) computed tomography scans. However, recent developments and studies using 4D imaging as a diagnostic approach have been promising. A recent statistical study by Giessen et al. [4] was developed to detect wrist abnormalities based on statistical model of wrist motion pattern and bone shape, and this study was performed using 4D-RX system. In another study by Berdia et al. [5], it was found that carpal bones exhibit hysteresis during wrist motion which was observed in the difference in carpal motion path of a normal bi-directional wrist motion. This means that the instantaneous position of carpal bones is not only dependent on the position of the wrist but also on the direction in which the motion is occurring. Hence, step-andshoot methods of imaging will not be able to detect and measure this hysteresis effect - the magnitude of which may be used to quantify carpal instability patterns. For this study, in order for the motion pattern of the geometric phantom to be similar to the wrist motion, the motion amplitudes between 5 mm to 20 mm were chosen. This is based on the largest motion amplitude within the proximal carpal row, during radioulnar deviation (20 ulnar deviation to 10 radial deviation), which occurs at the distal scaphoid and averages 12.4 mm [6]. The feasibility of 4D CT imaging was rst shown in our previous study by Tay et al. [1] using a 64-slice SSCT scanner with a temporal resolution of 165 ms. However, it required factory modications of the scanner to perform scans at a table speed slower than in the commercially available protocols, i.e. with a pitch of 0.1. This study used a 64-slice dual source computed tomography (DSCT) scanner (SOMATOM Denition by Siemens Healthcare, Forchheim, Germany) which has an improved temporal resolution of 83 ms [7]. This promises equivalent dynamic image quality without requiring any modications to the scanning pitch parameter. The purpose of this study is to assess the quality of 4D CT images of a periodically moving geometric phantom in a DSCT scanner with a clinical scan protocol at a pitch of 0.2.

Fig. 1. High-resolution phantom with targets at three planes. (Inset (Top): Orientation of the axial, sagittal and coronal planes. Inset (Bottom): A typical instantaneous velocity of moving phantom over one motion period of 1.5 s).

Data were acquired while the phantom was at rest (using the ECG signal simulated by the CT system for a spiral static protocol) and during periodic linear motion along the x-axis at a frequency of 40 cpm with amplitudes ranging from 5 mm to 20 mm. The corresponding maximum instantaneous velocities (hereafter termed velocities) of the phantom moving at a frequency of 40 cpm with amplitudes of 5 mm to 20 mm along the x-axis are shown in Table 1. In addition to this, the motion frequency was calculated based on Ohnesorge et al. study [8], where the maximum pitch value is given by pitch M 1 Trot M Tp (1)

2. Materials and methods 2.1. Equipment set-up A 64-slice DSCT scanner (SOMATOM Denition, Siemens Healthcare, Forchheim, Germany) was used to acquire all images with a retrospectively gated CT protocol at 0.2 pitch value, 60 mAs/rot, 120 kV and 0.33 s rotation time. A geometric spatial resolution phantom (QRM, Mhrendorf, Germany) with resolution targets in the axial (xy), coronal (xz) and sagittal (yz) planes was mounted onto a linear slider on a custom made motion platform powered by a servo motor (Mitsubishi Electric Corporation, Tokyo, Japan) (Fig. 1). The phantom was set in periodic linear motion along the x-axis slider by the servo motor system, which was controlled using TwinCAT I/O real time driver software (Beckhoff Automation GmbH, Verl, Germany). A customized data logging software was used to record the instantaneous velocities of the moving phantom (data was sampled every 0.12 s) with respect to its corresponding displacement on the x-axis slider (Fig. 1). The motion prole presented was chosen to give the maximum duration of time at a known constant velocity. A simulated electrocardiogram (ECG) signal in the form of a 10 ms voltage spike was generated at the end of each motion cycle to simulate the R wave of the QRS complex typically seen in a cardiac cycle.

where M is the number of detector rows used in the gated mode. Hence, for 0.2 pitch value, using Eq. (1), the minimum motion frequency is 37.5 cpm. Thus, the motion frequency of 40 cpm was selected to satisfy the minimum frequency requirement for a scan at pitch value of 0.2 to avoid image interpolation errors [8,9]. 2.2. Image reconstruction For each scan, data sets were reconstructed at every one tenth of a period (10 ECG phases) with a slice width of 0.6 mm, 0.6 mm reconstruction increment, 200 mm eld of view, 512 512 pixel and medium-sharp reconstruction kernel (B46f). The number of phases chosen was sufcient for this study, as we were able to reconstruct a smooth 3D movie running at around 7 frames per second (fps). 2.3. Image assessment The acrylic 3D spatial resolution phantom (QRM-3DSR-02, Moehrendorf, Germany) used in this study is made up with a series of drilled holes with varying diameter and spacing from 0.4 mm to 4.0 mm, allowing for an order of magnitude in spatial frequency. As the holes within the phantom move along the x-axis, projections acquired at different angular positions see the holes at different locations along the axis. The image slice corresponding to the middle of each resolution target was used to assess spatial resolution and image artifacts. Quantitative measurements were performed using Analyze 8.1 software (Mayo Foundation for Medical Education and Research, Rochester, MN).

P.Y. Neo et al. / Medical Engineering & Physics 35 (2013) 18371842 Table 1 Maximum instantaneous velocities of phantom moving at 40cpm with amplitudes 5 mm20 mm. Frequency (cpm) Amplitude (mm) Maximum Instantaneous Velocity (mm/s) 40 5 9.0 40 7 12.5 40 8 14.5 40 9 16.5 40 10 18.0 40 12 22.0 40 15 27.0

1839

40 20 36.0

Spatial resolution was quantied by the minimum diameter of holes that could be resolved on all 3 planes [1]. All 10 phases of motion were assessed, and the average spatial resolution was calculated. Banding artifacts [10] were observed only in images of the coronal resolution target, and were quantied by measuring the maximum displacement found between bands using the line prole tool in Analyze 8.1. All 10 phases were assessed, and the mean was calculated. Motion artifacts observed in the axial resolution target were scored on a 4-point scale [1]. All 10 phases were assessed, and the mean artifact score was calculated. Motion artifacts in the coronal and sagittal resolution targets were assessed by the length of the shading artifact (from the edge of the target) along the x-axis and y-axis respectively [1]. Using the intensity prole in the line prole tool in Analyze 8.1, the number of voxels it takes for the CT numbers to drop from 0 HU (Hounseld Units) to 900 HU was measured and recorded. This range was chosen to make allowances for the actual CT numbers for acrylic and air which were 100 HU and 1000 HU respectively. A shading length of less than 3 mm was considered mild, 36 mm was moderate, and severe for length of more than 6 mm. All 10 phases were assessed, and the mean was tabulated. Relationships between image quality and motion velocity were determined by linear regression. (Analysis Toolpak, Excel 2010, Microsoft Inc, Washington, USA)

3.3. Motion artifacts 3.3.1. Axial The motion artifact score (averaged over 10 phases) for the axial resolution target is shown in Fig. 3. Mean axial motion artifact was mild at velocities up to 18 mm/s, and moderate at velocities up to 36 mm/s. A relationship (motion artifact score = 0.08[s/mm] velocity[mm/s], p < 0.001) between axial motion artifact and velocity was determined by linear regression. 3.3.2. Coronal In the coronal resolution target, motion artifacts were classied by the magnitude of the shading effect in the x-axis. The mean shading length (averaged over 10 phases) as a function of velocity is shown in Fig. 4. Motion artifacts in the coronal plane were minimal for velocities up to 36 mm/s. A relationship (mean shading length = 0.04[s] velocity[mm/s], p < 0.001) between coronal motion artifact and velocity was determined by linear regression. 3.3.3. Sagittal In the sagittal resolution target, motion artifacts were classied by the magnitude of the shading effect in the y-axis. The mean shading length (averaged over 10 phases) as a function of velocity is shown in Fig. 5. Motion artifacts in the sagittal plane were minimal at velocities up to 36 mm/s. A relationship (mean shading length = 0.02[s] velocity[mm/s], p < 0.05) between sagittal motion artifact and velocity was determined by linear regression. 4. Discussion

3. Results 3.1. Spatial resolution The minimum diameter of holes that could be resolved in the three resolution targets of the moving phantom averaged for 10 phases is shown in Table 2. Sub-millimeter spatial resolution was achieved in all three planes for velocities of up to 36 mm/s. Spatial resolution in the sagittal plane was excellent and independent of motion velocity. Slight degradation in spatial resolution was observed in the coronal plane as motion velocities increased to 18 mm/s. Spatial resolution in the axial plane was most affected by motion velocity and showed a consistent degradation in spatial resolution with increasing motion velocity. The maximum measured spatial resolution within a motion cycle is presented in Table 2. A relationship (Axial spatial resolution = 0.67 mm + 0.01 s velocity[mm/s], p < 0.001) between spatial resolution at the axial plane and velocity was determined by linear regression. During radioulnar deviation (20 ulnar deviation to 10 radial deviation), the largest motion amplitude within the proximal carpal row occurs at the distal scaphoid and averages 12.4 mm [6]. Motion amplitudes between 5 mm and 20 mm were thus chosen for this study. Using the described motion simulator set up, this correspondingly gave maximum velocities between 9 mm/s and 36 mm/s. At a xed motion frequency of 40 cpm, this in turn translates to mean motion velocities of between 6.67 mm/s and 26.7 mm/s over the entire cycle including acceleration and deceleration at the end of the extremes of the range of the motion. The fundamental design of the motion simulator is similar to the simulator used in the study by Tay et al. [1]. We have assessed our results based on the maximum instantaneous velocities of the phantom even though the image quality results were averaged over all 10 phases of phantom motion. It is important to note that the initial and ending phases of phantom motion have better image qualities than the mid-phases due to a much slower motion when the phantom decelerates and changes direction at the motion extremities. The average image quality of the 10 motion phases was determined with respect to the maximum instantaneous velocity attained for each 4D CT scan at different motion amplitudes. This approach is similar to a study by Tay et al. [1]. Our results showed that sub-millimeter resolution of dynamically moving objects with largely minimal amounts of artifacts is achievable with current clinical DSCT scanners for motion amplitudes up to 20 mm using the scan and motion protocols detailed. This motion amplitude translates to a mean motion velocity of

3.2. Banding artifacts Banding artifacts were only present in the coronal plane. They were minimal for velocities up to 18 mm/s and mild for velocities up to 36 mm/s, with the maximum shift of 1.48 mm. A relationship describing the maximum band shift averaged over 10 phases (hereafter termed mean band artifact) between bands in the x-axis direction as a function of velocity (mean band artifact = 0.04[s/mm] velocity[mm/s], p < 0.001) was determined by linear regression and is shown in Fig. 2.

1840 Table 2 Mean spatial resolution as a function of velocity. Frequency (cpm) Amplitude (mm) Velocity (mm/s) Axial Plane Coronal Plane Sagittal Plane Mean over three planes 0 0 0 0.7 (0.7) 0.7 (0.7) 0.7 (0.7) 0.7 40

P.Y. Neo et al. / Medical Engineering & Physics 35 (2013) 18371842

40 7 12.5 0.74 (0.8) 0.71 (0.9) 0.7 (0.7) 0.72

40 8 14.5 0.78 (0.9) 0.7 (0.9) 0.73 (0.8) 0.74

40 9 16.5 0.77 (1.0) 0.7 (0.9) 0.73 (0.9) 0.73

40 10 18.0 0.87 (1.2) 0.8 (1.0) 0.7 (0.8) 0.79

40 12 22.0 0.88 (2) 0.8 (1.0) 0.7 (0.8) 0.79

40 15 27.0 0.93 (1.5) 0.8 (0.9) 0.7 (0.7) 0.81

40 20 36.0 0.96 (3.0) 0.8 (1.0) 0.7 (0.7) 0.82

5 9.0 0.74 (0.8) 0.75 (0.9) 0.7 (0.7) 0.73

* (Value) is the maximum measured spatial resolution in a motion cycle.

Fig. 2. Maximum band shift averaged over 10 phases as a function of velocity.

26.7 mm/s for our set up. Motion artifacts at motion amplitude of 20 mm were mild in the coronal and sagittal planes but moderate in the axial plane. This implies that the orientation of the moving object in a dynamic scan should have their primary motion in the coronal and sagittal plane for the best image quality. We have analyzed the axial artifacts using a subjective scoring scale that has been published [1]. Due to the complexity of the motion artifacts, there is no known quantitative way to measure it. We acknowledge this as a limitation.

The cause to the degradation of the resolution is mainly due to motion blur. Motion blur becomes more pronounced as the velocity increases. Thus it is expected to nd a linear relationship between velocity and resolution. Similarly, the severity of banding artifacts around the image is dependent on velocity. A recent study by Tang et al. [11] resolved the scaphoid movement during radioulnar deviation into three planes: (1) the exion/extension plane, (2) the supination/pronation plane and (3) the radioulnar deviation plane, and reported that the

Fig. 3. Axial motion artifact score averaged over 10 phases as a function of velocity.

P.Y. Neo et al. / Medical Engineering & Physics 35 (2013) 18371842

1841

Fig. 4. Coronal plane shading length averaged over 10 phases as a function of velocity.

largest motion amplitude of the scaphoid happens in the exion/extension plane. Given the results from this study, this would mean that for future cadaveric or in vivo studies of wrist radioulnar deviation, we should consider positioning the exion/extension plane of the scaphoid distal pole in the coronal or sagittal planes. Another limitation of this study was in not evaluating the effect of motion along the longitudinal z-axis. We did not analyze the zdisplacement as we felt that the displacement along the z-axis, is considerably less than the displacement along the x-axis. In future validation studies, we will analyze image quality in all 3-axis of motion. However, as there are other parameters affecting the nal overall quality of the images, further studies on image quality would be needed to come to a more denite conclusion on the best forearm orientation to adopt for dynamic radioulnar deviation imaging. Briey, these parameters include the orientation of

non-isotropic CT image voxels and the effect of the banding artifacts which will always appear perpendicular to the CT scanner gantry plane. We also compared our results with those reported by Tay et al. using a similar set-up [1]. In that study, a single-source CT scanner with a non-clinical pitch of 0.1 was used. However, as the study did not have a sagittal plane target, our comparison are limited to just the axial and coronal planes. Spatial resolution at the axial and coronal planes were comparable between both studies, with both the current study and the previous study reporting sub-millimeter resolution at velocities up to 36 mm/s and 40 mm/s respectively. Though the spatial resolution for SSCT was expected to be more affected by motion blur, their results were comparable due to the reconstruction kernel chosen. In SSCT study, the experiment was done using a modied research CT scanner with a sharp kernel of B70. In this study, it was done using clinical CT scanner with a kernel of B46f. Thus, as the choice of reconstruction kernel can

Fig. 5. Sagittal plane shading length averaged over 10 phases as a function of velocity.

1842

P.Y. Neo et al. / Medical Engineering & Physics 35 (2013) 18371842

affect the image quality as well, both studies were able to achieve comparable, sub-millimeter resolution. We also noted that measuring point spread function (PSF) is another method to assess the spatial resolution. It might be possible that PSF measurement would show a poorer resolution. However, in this study, we nd that analyzing the resolution subjectively based on human ability to resolve the hole separately is more practical and clinically relevant. Banding artifacts of the dynamic images were caused by the malalignment of the reconstructed, image of the object scanned along the coronal plane (xz axis), where the neighboring bands corresponding to consecutive motions was shifted with respect to each other along the x-axis [1]. This was the result of parallax errors, in which there was a time difference between the images captured at each angular position. Banding artifacts in the coronal plane were present in both studies. However, the current study showed marked improvement in the magnitude of the banding artifact with respect to velocity with a current relationship of (mean band artifact = 0.04[s/mm] velocity[mm/s], p < 0.001) compared to the previously reported relationship (mean band shift = 0.13[s/mm] velocity[mm/s], p < 0.0001). The magnitude of the motion artifact in the axial plane, although slightly larger than the previous study, has the same classication of severity in both studies. In the coronal plane, motion artifact manifested by the length of the shading effect was smaller than in the previous study, with mean shading lengths almost 5 times shorter at any given velocity. The overall improved image quality in the current study is consistent with the different temporal resolution of the CT scanners used in the previous study [1], 83 ms and 165 ms respectively. While the lowest pitch that is available determines how slow the phantom or joint can be moved, it should not have major impact on image quality. It would also be benecial to explore the quality of dynamic images using a CT with a greater number of scanning slices, which might widen the scope of dynamic imaging to include joints larger than the wrist. Given that image quality is dependent on motion velocity, decreasing the motion velocity or using smaller motion amplitude would further improve image quality. The former is controlled by the frequency of motion (a lower motion frequency would be needed) and the lowest possible frequency that can be used for a scanning pitch of 0.2 would be 37 cpm. Given that motion amplitude of 20 mm is sufciently larger than the reported average maximum displacement of the scaphoid in the proximal carpal row, we did not explore the image quality of motion amplitudes larger than 20 mm. However, it would be relevant for future studies to investigate the upper amplitude threshold for dynamic studies of larger joints such as the elbow or knee. In the study by Tang et al. [11], it was also mentioned that the greatest scaphoid displacement during radioulnar deviation occurs during slight ulnar deviation, while the smallest scaphoid displacement occurs during radial deviation. Further work to correlate the amount of radial and ulnar deviation of an actual hand to the amount of distal scaphoid displacement would enable future clinical applications of the current motion simulator. A recent paper by Leng et al. [12] described the use of a dynamic technique to capture motion of a moving carpal joint without the need for periodic motion. By just requiring a single cycle of motion, the technique described by Leng et al. would result in a lower radiation dose compared to our technique which we acknowledge is less advantageous. However, note should be taken that (i) the wrist is

not a radiation sensitive organ and (ii) the eld of view in Leng et als study was limited and only the carpus could be visualized. The latter would result in difculties quantifying carpal hysteresis using the technique reported by Berdia et al. [5], given that the motion of each of the carpal bones has to be tabulated with respect to the third metacarpal motion. Using a clinical DSCT scanner, we have determined the boundary parameters of which sub-millimeter resolution images with predominantly minimal to mild motion and banding artifacts can be obtained from a moving target; thereby demonstrating the feasibility of using DSCT scanners at clinical scanning parameters for 4D CT wrist imaging of the proximal carpal row. Competing interests None declared. Funding This research is supported by the Singapore Ministry of Healths National Medical Research Council under its Individual Research Grant (Grant no: NMRC/1148/2007). Ethical approval Not required. Acknowledgements The authors thank radiographers Hong WL and Mahmood B NI for their technical assistance, Andrew Primak (PhD) and Liu Xuan for their invaluable inputs to the manuscript. References
[1] Tay SC, Primak AN, Fletcher JG, Schmidt B, An KN, McCollough CH, et al. Understanding the relationship between image quality and motion velocity in gated computed tomography: preliminary work for 4-dimensional musculoskeletal imaging. J Comput Assist Tomogr 2008;32(4):6349. [2] Carelsen B, Jonges R, Strackee SD, Maas M, van Kemenade P, Grimbergen Ca van Herk M, et al. Detection of in vivo dynamic 3-D motion patterns in the wrist joint. IEEE Trans Biomed Eng 2009;56(4):123644. [3] Founami M, Strackee SD, Jonges R, Blankevoort L, Zwinderman AH, Carelsen B, et al. In-vivo three-dimensional carpal bone kinematics during exion-extension and radio-ulnar deviation of the wrist: dynamic motion versus step-wise static wrist positions. J Biomech 2009;42(16):2664 3267. [4] van de Giessen M, Founami M, Vos FM, Strackee Sd Maas M, Van Vliet LJ, Grimbergen CA. Streekstra GJ A 4D statistical model of wrist bone motion patterns. IEEE Trans Med Imag 2012;31(3):61325. [5] Berdia S, Short WH, Werner FW, Green JK, Panjabi M. The hysteresis effect in carpal kinematics. J Hand Surg [Am] 2006;31(4):594600. [6] Tay SC, Primak AN, Fletcher JG, Schmidt B, Amrami KK, Berger RA, et al. Fourdimensional computed tomographic imaging in the wrist: proof of feasibility in a cadaveric model. Skeletal Radiol 2007;36(12):11639. [7] Petersilka M, Bruder H, Krauss B, Stierstorfer K, Flohr TG. Technical principles of dual source CT. Eur J Radiol 2008;68(3):3628. [8] Ohnesorge B, Flohr T, Becker C, Kopp AF, Schoepf UJ, Baum U, et al. Cardiac imaging by means of electrocardiographically gated multisection spiral CT: initial experience. Radiology 2000;217(2):56471. [9] Kang JW, Do KH, Chung JY, Cho HJ, Seo JB, Lim TH, et al. Concept of minimal heart rate for each pitch value to avoid interpolation artifact when using dual-source CT: a phantom study. Int J Cardiovasc Imaging 2010;26(Suppl 1):1039. [10] Taguchi K, Chiang BS, Hein IA. Direct cone-beam cardiac reconstruction algorithm with cardiac banding artifact correction. Med Phys 2006;33(2):52139. [11] Tang JB, Xu J, Xie RG. Scaphoid and lunate movement in different ranges of carpal radioulnar deviation. J Hand Surg [Am] 2011;36(1):2530. [12] Leng S, Zhao K, Qu M, An KN, Berger R, McCollough CH, et al. Dynamic CT technique for assessment of wrist joint instabilities. Med Phys 2011;38(Suppl. 1):S50.