Dose Reduction in Helical CT: Dynamically Adjustable z-axis X-Ray Beam Collimation

Transcription

1 Medical Physics and Informatics Original Research Christner et al. CT Dose Reduction Medical Physics and Informatics Original Research Downloaded from by on 2/26/18 from IP address Copyright ARRS. For personal use only; all rights reserved Jodie A. Christner 1 Vanessa A. Zavaletta 1 Christian D. Eusemann 1,2 Alisa I. Walz-Flannigan 1 Cynthia H. McCollough 1 Christner JA, Zavaletta VA, Eusemann CD, Walz-Flannigan AI, McCollough CH Keywords: CT, dose efficiency, helical CT, overscanning, radiation dose DOI:1.2214/AJR Received April 8, 29; accepted after revision June 26, 29. The project described was supported by grant DK837 from the National Institutes of Health. The content is solely the responsibility of the authors and does not necessarily represent the official view of the National Institutes of Health. Additional support was provided by a research grant from Siemens Healthcare. The employment status of C. D. Eusemann at Siemens Healthcare did not influence the data in this study. 1 Department of Radiology, Mayo Clinic Rochester, 2 First St. SW, East-2 Mayo Bldg., Rochester, MN Address correspondence to C. H. McCollough (mccollough.cynthia@mayo.edu). 2 Siemens Healthcare, Malvern, PA. WEB This is a Web exclusive article. AJR 21; 194:W49 W X/1/1941 W49 American Roentgen Ray Society Dose Reduction in Helical CT: Dynamically Adjustable z-axis X-Ray Beam Collimation OBJECTIVE. The purpose of this study was to measure the dose reduction achieved with dynamically adjustable z-axis collimation. MATERIALS AND METHODS. A commercial CT system was used to acquire CT scans with and without dynamic z-axis collimation. Dose reduction was measured as a function of pitch, scan length, and position for total incident radiation in air at isocenter, accumulated dose to the center of the scan volume, and accumulated dose to a point at varying distances from a scan volume of fixed length. Image noise was measured at the beginning and center of the scan. RESULTS. The reduction in total incident radiation in air at isocenter varied between 27% and 3% (pitch,.5) and 46% and 8% (pitch, 1.5) for scan lengths of 2 and 5 mm, respectively. Reductions in accumulated dose to the center of the scan were 15% and 29% for pitches of.5 and 1.5 for 2-mm scans. For scan lengths greater than 3 mm, dose savings were less than 3% for all pitches. Dose reductions 8 mm or farther from a 1-mm scan range were 15% and 4% for pitches of.5 and 1.5. With dynamic z-axis collimation, noise at the extremes of a helical scan was unchanged relative to noise at the center. Estimated reductions in effective dose were 16% (.4 msv) for the head, 1% (.8 and 1.4 msv) for the chest and liver, 6% (.8 msv) for the abdomen and pelvis, and 4% (.4 msv) and 55% (1. msv) for coronary CT angiography at pitches of.2 and 3.4. CONCLUSION. Use of dynamic z-axis collimation reduces dose in helical CT by minimizing overscanning. Percentage dose reductions are larger for shorter scan lengths and greater pitch values. C onventional helical CT techniques require irradiation of anatomic features outside the desired reconstruction volume to reconstruct images at the start and end of the scan range (Fig. 1), an effect referred to as overscanning [1 5] or overranging [6]. Patient dose increases as the x-ray beam and detector widths are increased, and relative dose increases as the scan range is decreased [1 3, 6]. Previous estimates of the percentage dose increase due to overscanning relative to contiguous nonhelical acquisitions at the same beam collimation and desired reconstruction volume varied from 13% to 36% for a 16-MDCT system with a total nominal beam width of 24 mm [3]. Increasing pitch from.7 to 1.4 was found to increase overscanning as much as a factor of 2 for a similar 16-MDCT scanner [6]. However, a variety of behaviors for overscanning have been observed and not all were linearly dependent upon pitch [6]. Overscanning is required in helical CT for acquisition of sufficient projection data on each side of the desired reconstruction volume [7]. To reduce overscanning and improve dose efficiency, Tang et al. [8] proposed a raywise 3D weighted cone-beam filtered back-projection algorithm to extend the reconstructible region allowed by conventional reconstruction algorithms and thereby substantially reduced the requirement for overscanning. However, image noise increased at the extremes of the scan range. Alternatively, Stierstorfer et al. [9] proposed a hardware-based solution whereby x-ray beam collimation is increased from zero to the total nominal beam width as acquisition progresses from the start position to one beam width in from the start position. As the scan position reaches one beam width from the end location, the collimation opening is decreased, reaching zero at the end location. Implementation of this approach requires AJR:194, January 21 W49

2 Christner et al. Downloaded from by on 2/26/18 from IP address Copyright ARRS. For personal use only; all rights reserved Normalized Dose (%) Normalized Dose (%) Overscan Regions Relative z-axis Position (mm) moving collimator jaws capable of velocities up to 35 mm/s for a table speed of up to 192 mm/s (focus to isocenter distance, 595 mm). This approach has been implemented in a commercially available CT system. The purpose of this study was to experimentally determine the dose reduction achieved with this technique and to assess image quality at the extremes of the scan range. Dose reductions were determined for the following three conditions: total radiation output in air at isocenter, dose to a point as a function of scan length for scans centered over that region (e.g., breast dose for pulmonary emboli scans of varying z-axis coverage but centered over the breast [1]), and dose to a point as a function of distance from the center of the scan range for scans of a fixed length (e.g., breast dose from 1-mm scans centered on the head, chest, or abdomen). Materials and Methods CT System All data were acquired in helical mode on a single-source 64-MDCT system (Somatom Definition AS+, Siemens Healthcare) equipped with dynamic z collimation. A total nominal beam width of 38.4 mm and rapid oscillations of the focal spot position (z-flying focal spot) were used to acquire Overscan Regions Relative z-axis Position (mm) A C Normalized Dose (%) 12 1 unique projections along the z-axis for every view angle (i.e., mm acquisition mode). Data were collected and compared for helical acquisitions with fixed z-axis beam collimation or dynamic z collimation (Adaptive Dose Shield, Siemens Healthcare), all other parameters being the same. Fixed z-axis beam collimation was accomplished with the assistance of the manufacturer. Fig. 2 Photograph shows phantom setup used to determine (1) dose reduction to a specific point centered within the scan volume, as a function of length of scan volume, and (2) dose reduction to a point as a function of the distance from that point to the center of the scan volume for 1-mm-long scan range. Arrow shows position of sensor within phantom No Overscan Relative z-axis Position (mm) Fig. 1 Graphs illustrate z-axis dose profiles (solid lines) of overscan regions for 1-mm scan length. A, Fixed z collimation; detector length, 4 mm; pitch, 1; slice length, 1 mm. B, Same as A except dynamic x-ray beam collimation restricts dose to length of scan volume. C, Same as A except pitch is 1.5. Visual Assessment of Irradiated Region To show the width of the irradiated region as a function of time during a helical scan, an intensifying screen from an x-ray cassette (Kodak Lanex Fast B, Carestream Health) was irradiated during helical acquisitions of 8 seconds duration with both fixed and dynamic z collimation. A digital movie camera was used to record the luminescent response of the screen, and digital frames were extracted at 1-second intervals during the scan. Total Radiation Output in Air at Isocenter The total radiation output for different scan lengths with or without dynamic z collimation was determined as the CT dose index 1 (CTDI 1 ), B W5 AJR:194, January 21

3 CT Dose Reduction Downloaded from by on 2/26/18 from IP address Copyright ARRS. For personal use only; all rights reserved expressed as absorbed dose to air according to the guidelines published by the European Commission [11]. A 1-mm-long, 3-cm 3 volume CTDI ionization chamber (model CT, Radcal) and associated dosimeter (model 915, Radcal) were used. The chamber was centered at scanner isocenter with its length parallel to the z-axis. An external support mechanism was used such that the chamber did not move during table translation, and the table remained outside the x-ray beam. The scan parameters were 12 kv, 2 effective mas, mm collimation,.5-second rotation time, and pitches of.5, 1., and 1.5. The scan length varied from 2 to 5 mm. Dose to a Point Within a Phantom Dose at the center of the scan volume for scan lengths from 2 to 9 mm and dose to a point 23 mm from the center of a 1-mm scan length were measured with a solid-state point detector and associated multimeter (Barracuda detector type CT-SD16 version 1.1, RTI Electronics). The point detector was placed at the center of a 1.25-m-long acrylic phantom (Fig. 2) consisting of a human skull embedded in acrylic and several solid acrylic cylinders of diameters varying from 14 to 32 mm. The scan parameters were 12 kv, 2 effective mas, collimation,.5-second rotation time, and pitches of.5, 1, and 1.5. Tube current modulation was not enabled. To measure the accumulated dose to a point located at varying distances from the scan region, the center of the scan volume was moved in relation to the detector, instead of vice versa, to keep the detector at the same position within the phantom. Percentage dose reduction was calculated as the difference in dose measured with a standard fixed z collimation (D Sz ) versus dose measured with dynamic z collimation (D Dz ) normalized to the dose measured with a standard fixed z collimator, as follows: Percentage dose reduction = D Sz D Dz D Sz Image Quality CT attenuation and noise were measured using a cylindric water calibration phantom (2 mm in diameter, 1 mm thick) positioned at isocenter. The scans were acquired with dynamic z collimation, 12 kv, 2 effective mas (no tube current modulation),.5-second rotation time, and pitch of 1. Images were reconstructed with a slice thickness of.6 mm, 22-mm field of view, and a medium sharp reconstruction kernel (B4). Noise was measured in an image corresponding to the center of the water phantom with a scan length of 1 mm. Three scans were acquired with the scan volume centered at the phantom center. An additional three scans were acquired with the scan volume starting at the center of the phantom. Circular regions of interest 11 mm in diameter were placed at the center and the 12, 3, 6, and 9 o clock positions. The peripheral regions of interest were centered 11 mm from the phantom edge. The mean and SD of the attenuation coefficient values within the regions of interest were recorded as CT attenuation and noise, respectively. Clinical Importance Reductions in effective dose with dynamic z collimation were estimated for protocols in use at our institution for CT examinations of the head, chest (pulmonary embolism), coronary arteries (CT angiography), liver (arterial phase of a triphasic examination), and abdomen and pelvis. For coronary CT angiography, pitch and effective dose were appropriate for a heart rate of 6 beats/min; ECG-based tube current reduction to achieve 3% dose reduction was used. To compute effective dose, the dose length products displayed by the scanner were multiplied by published effective dose per dose length product factors [12]. The percentage reduction in total incident radiation in air at isocenter as a function of pitch and scan length was used to calculate reductions in effective dose because effective dose scales linearly with both total scan time and dose length product. Dose reductions for coronary CT angiography at pitches of.2 and 3.4 were estimated by extrapolation of a linear fit (r 2 =.98) to dose reductions measured for pitch values of.5, 1., and 1.5 and for a scan length of 12 mm. Results Visual Assessment of Irradiated Region The opening and closing of the collimator as a function of time were seen on the luminescent screen images acquired with and without dynamic z collimation (Fig. 3). Total Radiation Output in Air at Isocenter The mean total reduction in CTDI 1 in air at isocenter (Table 1) was constant for the scan lengths studied and was 12.7 ±.3 (SD), 31.3 ±.6, and 5.6 ± 1.4 mgy for pitches of.5, 1., and 1.5. This reduction represents the extra radiation generated during overscanning and avoided with use of the dynamic z collimator. The maximum percentage reduction in CTDI 1 in air varied between 27% and 46% for 2-mm scans with pitches of.5 and 1.5 (Fig. 4). For scan lengths of 5 mm, dose reduction varied between 3% and 8% for the same pitches. Because a fixed amount of radiation was avoided, percentage dose reduction due to dynamic z collimation decreased as scan length increased. Hence, reductions were greater for shorter scans and higher pitches. These data represented the reduction in primary radiation emitted during the scan. Dose to a Point Within a Phantom Conversely, the dose reduction in the phantom at the center of the scan volume as a function of scan length was a reduction in scattered radiation because the measurement TABLE 1: Absolute Reduction in CT Dose Index 1 (CTDI 1 ) in Air at Isocenter as Function of and Pitch Integrated Over Total Scan Time (mm) Dose Reduction (mgy) Pitch.5 Pitch 1. Pitch Mean SD Note CTDI 1 in air at isocenter (i.e., with no phantom attenuation) was 19.7 ±.2 mgy/1 mas for one axial scan with the same beam collimation and technique factors. AJR:194, January 21 W51

4 Christner et al. Downloaded from by on 2/26/18 from IP address Copyright ARRS. For personal use only; all rights reserved A B C D E F G H I J K L M N O P Q R S T U V Fig. 3 Image frames captured from movies of luminescent screen. A K, Fixed collimation at (A), 1 (B), 2 (C), 3 (D), 4 (E), 5 (F), 6 (G), 7 (H), 8 (I), 9 (J), and 1 seconds (K). L V, Dynamic z collimation at (L), 1 (M), 2 (N), 3 (O), 4 (P), 5 (Q), 6 (R), 7 (S), 8 (T), 9 (U), and 1 seconds (V). W52 AJR:194, January 21

5 CT Dose Reduction Downloaded from by on 2/26/18 from IP address Copyright ARRS. For personal use only; all rights reserved Dose Reduction (%) (mm) Fig. 4 Graph shows percentage dose reduction in total radiation output (CT dose index 1) as function of scan length. Diamonds indicate pitch of.5; squares, pitch of 1.; triangles, pitch of 1.5. Dose Reduction (%) (mm) point was within the scan volume (i.e., primary radiation field) for all scan lengths. The dose reductions were greatest for the shortest scan lengths and highest pitch values (Table 2, Fig. 5) and were 1.1, 1.8, and 3.4 mgy (15%, 2%, and 28%) for pitches of.5, 1., and 1.5 for a scan length of 2 mm. For scans longer than 3 mm, minimal dose reduction ( 3%) was found for all pitch values. Because the distance between the scan volume and dose measurement location varied, the dose reduction to a specific point for a 1-mm scan length reflected a reduction in both scattered and primary radiation because some measurement points were located within the scan volume and some were outside the scan volume. The maximum absolute values in dose reduction for these conditions (Table 3) occurred when the dosimeter was approximately 8 mm from the start or end of the scan and were.9, 2.5, and 4.2 mgy for pitches of.5, 1., and 1.5. When the detector was positioned within the scan region (between and < 5 mm in Fig. 6), , Fig. 5 Graph shows percentage dose reduction to specific point centered within the scan volume as function of length of scan volume. Diamonds indicate pitch of.5; squares, pitch of 1.; triangles, pitch of dose reductions were 8%, 16%, and 2% for pitch values of.5, 1., and 1.5. When the detector was positioned outside the scan region (> 5 mm distance in Fig. 6), dose reductions were 15%, 3%, and 4% for pitch values of.5, 1., and 1.5. However, beyond 5 mm from the dose measurement point, the absolute dose levels, which are due to scattered radiation only, were very small. Hence, a 4% dose reduction reflected only a small change in absolute dose (Table 3). Image Quality When the scan was centered on the water phantom, the image noise measured in the central image of the phantom was 16.4 ±.2 HU. When the first image of the prescribed scan range was centered on the water phantom, image noise was 16.3 ±.2 HU. At all four of the peripheral positions, the mean TABLE 2: Absolute Dose Reduction to a Point in the Phantom as a Function of and Pitch for Scans Centered on the Measurement Position (mm) Dose Reduction (mgy) Pitch.5 Pitch 1. Pitch TABLE 3: Absolute Reduction in Accumulated Dose to a Point in the Phantom as a Function of Distance Between That Point and the 1-mm-Long Scan Range Distance to Scan Range (mm) Dose Reduction (mgy) Pitch.5 Pitch 1. Pitch noise values (Table 4) at the start of the scan and at the scan center were not significantly different from each other. The CT number of water ranged from 1.2 to.3 HU, well within the typical required limits of ± 5 HU. Clinical Importance The reductions in effective dose estimates for clinical examinations were 16% (.4 msv) for the head, 1% (.8 and 1.4 msv) for the chest and liver, and 6% (.8 msv) for the abdomen and pelvis (Table 5). For coronary CT angiography with the scanner studied, on which the pitch was approximately.2 for a heart rate of 6 beats/min, there was a 4% (.4 msv) dose reduction. However, extrapolation of the measurements to a pitch of 3.4, which is achievable with dualsource CT, yielded a 55% (1. msv) effective dose reduction. AJR:194, January 21 W53

6 Christner et al. Downloaded from by on 2/26/18 from IP address Copyright ARRS. For personal use only; all rights reserved Dose Reduction (%) Distance Between Dosimeter and Scan Region (mm) Discussion In previous work [2, 3, 6] the theoretic increase in total radiation output due to overscanning has been estimated with a trapezoidal model of the radiation dose profile along the z-axis. Use of a trapezoid model, together with the time series photos in Figure 3, can explain the measured dose saving. The time series photos show that with dynamic z collimation, collimator opening and closing occurred gradually, reducing the total radiation output compared with that of fixed z collimation. For fixed z collimation, the collimator was fully open in sequential frames of the movie. As the beam moved toward the start of the prescribed scan volume, the accumulated dose produced the slanted side of the trapezoid. For dynamic z collimation, however, the collimator was closed until the beam was at the beginning position of the scan volume. Then it opened only on the side entering the scan volume until it was fully opened. At the end of the scan, the leading edge of 5 Fig. 6 Graph shows percentage dose reduction to point as function of distance from that point to center of scan volume for 1-mm-long scan range. Diamonds indicate pitch of.5; squares, pitch of 1.; triangles, pitch of the collimator closed when it left the scan volume. Thus the dose profile became closer to rectangular and therefore more greatly resembled an axial (sequential) scan, in which there is no overscan region. In general, as pitch increases, the irradiation length for one 36 rotation must increase. Thus to reconstruct the first image in the scan volume, which requires 18 of data on either side of the image center, the irradiation length outside the scan volume must increase, increasing the amount of overscanning as pitch is increased. Use of the dynamic beam collimator, which opened and closed at a rate related to pitch, essentially eliminated overscanning and made the radiation dose profile independent of pitch. Thus as a percentage of the dose with fixed collimation, dose reduction achieved with dynamic z collimation was greater for higher pitch values. The percentage reduction in total radiation output, or dose, at the center point of a scan volume increased for shorter scan lengths and TABLE 5: Estimated Reductions in Effective Dose for Representative Clinical Protocols Anatomic Region Pitch (mm) CTDI vol (mgy) Dose Length Product (mgy cm) Effective Dose With Fixed z Collimation (msv) higher pitches. This means, for example, that the percentage dose reduction to the breast due to use of dynamic z collimation is greater for shorter than for longer scan acquisitions. Because noise at the beginning of the helical scan range was not significantly different from that at the center, the measured dose reduction was accomplished without sacrifice of image quality. This result, achieved with a hardware approach to solve the overscan problem, is in contrast to a previously described reconstruction algorithm approach [8] whereby image noise was increased at the extremes of the scan range. We conclude that in helical CT, dynamic z collimation can be used without sacrifice of image quality to reduce unnecessary dose due to overscanning. The percentage dose reductions are greatest for shorter scan lengths and greater pitch values. This technique may have greatest value in pediatric CT, in which smaller patient size requires smaller scan ranges and higher pitch values are used Reduction in Total Incident Radiation in Air at Isocenter With Dynamic z Collimation (%) Effective Dose Reduction With Dynamic z Collimation (msv) Head , Chest Liver Abdomen and pelvis Coronary arteries Single-source CT Dual-source CT Note All protocols have assumed collimation of mm and 12-kV tube potential. CTDI vol = volume CT dose index. TABLE 4: Image Noise With Dynamic z Collimation Region of Interest Clock Position Scan Center Image Noise (HU) First Image Center 16.4 ± ± ± ± ± ± ± ± ± ±.9 Note Tabulated values are mean ± SD (n = 3 for all measurements). Image noise is measured as SD of attenuation within region of interest. W54 AJR:194, January 21

7 CT Dose Reduction Downloaded from by on 2/26/18 from IP address Copyright ARRS. For personal use only; all rights reserved to minimize scan time. This technique also is attractive for cardiac CT, in which scan lengths are short ( 12 mm) and pitches greater than 3 are possible with dual-source scanners. Regardless of the magnitude of dose reduction, all helical scans will benefit from some level of dose reduction through the use of dynamic z collimation. Acknowledgments We thank Kristina Nunez for assistance with manuscript preparation and Tom Vrieze, Mike Bruesewitz, and David Holmes III for assistance with measurement techniques. References 1. Boone JM, Cooper VN 3rd, Nemzek WR, McGahan JP, Seibert JA. Monte Carlo assessment of computed tomography dose to tissue adjacent to the scanned volume. Med Phys 2; 27: Nicholson R, Fetherston S. Primary radiation outside the imaged volume of a multislice helical CT scan. Br J Radiol 22; 75: Tzedakis A, Damilakis J, Perisinakis K, Stratakis J, Gourtsoyiannis N. The effect of z overscanning on patient effective dose from multidetector helical computed tomography examinations. Med Phys 25; 32: Theocharopoulos N, Damilakis J, Perisinakis K, Gourtsoyiannis N. Energy imparted-based estimates of the effect of z overscanning on adult and pediatric patient effective doses from multi-slice computed tomography. Med Phys 27; 34: Tzedakis A, Damilakis J, Perisinakis K, Karantanas A, Karabekios S, Gourtsoyiannis N. Influence of z overscanning on normalized effective doses calculated for pediatric patients undergoing multidetector CT examinations. Med Phys 27; 34: van der Molen AJ, Geleijns J. Overranging in multisection CT: quantification and relative contribution to dose comparison of four 16-section CT scanners. Radiology 27; 242: Crawford CR, King KF. Computed tomography scanning with simultaneous patient translation. Med Phys 199; 17: Tang X, Hsieh J, Dong F, Fan J, Toth TL. Minimization of over-ranging in helical volumetric CT via hybrid cone beam image reconstruction: benefits in dose efficiency. Med Phys 28; 35: Stierstorfer K, Kuhn U, Wolf H, Petersilka M, Suess C, Flohr T. Principle and performance of a dynamic collimation technique for spiral CT. (abstr) In: Radiological Society of North America scientific assembly and annual meeting program. Oak Brook, IL: Radiological Society of North America, 27:SSA Parker MS, Hui FK, Camacho MA, Chung JK, Broga DW, Sethi NN. Female breast radiation exposure during CT pulmonary angiography. AJR 25; 185: European Commission. European guidelines on quality criteria for computed tomography (EUR EN). Luxembourg, Luxembourg: European Commission, Shrimpton P. Assessment of patient dose in CT. Appendix C, European guidelines for multislice computed tomography funded by the European Commission 24; contract number FIGM- CT2-278-CT-TIP. Luxembourg, Luxembourg: European Commission, 24 AJR:194, January 21 W55