Image quality and absorbed dose comparison of single‐ and dual‐source cone‐beam computed tomography

Abstract Purpose Dual‐source cone‐beam computed tomography (DCBCT) is currently available in the Vero4DRT image‐guided radiotherapy system. We evaluated the image quality and absorbed dose for DCBCT and compared the values with those for single‐source CBCT (SCBCT). Methods Image uniformity, Hounsfield unit (HU) linearity, image contrast, and spatial resolution were evaluated using a Catphan phantom. The rotation angle for acquiring SCBCT and DCBCT images is 215° and 115°, respectively. The image uniformity was calculated using measurements obtained at the center and four peripheral positions. The HUs of seven materials inserted into the phantom were measured to evaluate HU linearity and image contrast. The Catphan phantom was scanned with a conventional CT scanner to measure the reference HU for each material. The spatial resolution was calculated using high‐resolution pattern modules. Image quality was analyzed using ImageJ software ver. 1.49. The absorbed dose was measured using a 0.6‐cm3 ionization chamber with a 16‐cm‐diameter cylindrical phantom, at the center and four peripheral positions of the phantom, and calculated using weighted cone‐beam CT dose index (CBCTDI w). Results Compared with that of SCBCT, the image uniformity of DCBCT was slightly reduced. A strong linear correlation existed between the measured HU for DCBCT and the reference HU, although the linear regression slope was different from that of the reference HU. DCBCT had poorer image contrast than did SCBCT, particularly with a high‐contrast material. There was no significant difference between the spatial resolutions of SCBCT and DCBCT. The absorbed dose for DCBCT was higher than that for SCBCT, because in DCBCT, the two x‐ray projections overlap between 45° and 70°. Conclusions We found that the image quality was poorer and the absorbed dose was higher for DCBCT than for SCBCT in the Vero4DRT.

Vero4DRT (Mitsubishi Heavy Industries, Ltd., Hiroshima, Japan, and Brainlab, Munich, Germany) is a unique image-guided radiotherapy system comprising two imaging units aligned at AE45°relative to a megavoltage (MV) beam axis. Each imaging unit consists of a kV X ray tube and a FPD. Miura et al. 6 reported the image quality assurance (QA) for the Vero4DRT system with SCBCT. In SCBCT, it takes approximately 30 s to acquire the projection data using a 215°rotation because the rotation speed is limited to 7°/s. The Vero4DRT is now available with dual-source CBCT (DCBCT), in which it takes approximately 15 s to acquire the projection data using a 115°rotation, making it very useful for reducing the treatment time. In addition, DCBCT might reduce motion artifacts. DCBCT also plays a large role in some 4D-CBCT techniques, which may benefit from a dual-source technique. 9 Two important issues need to be addressed when using DCBCT. First, how does the image quality of DCBCT compare with that of SCBCT? Second, does DCBCT increase the patient dose compared with SCBCT? The purpose of CBCT is to provide a volumetric image for patient positioning for radiotherapy; thus, it is important to study the dose-image quality tradeoffs. However, no information is available on the imaging performance of a commercial DCBCT.
In this study, we evaluated the performance of a DCBCT, by comparing it with a SCBCT. Both were used in a Vero4DRT. The American Association of Physicists in Medicine (AAPM) Task Group (TG) 142 recommends several QA values for imaging by IGRT. 10 We focused on image uniformity, Hounsfield unit (HU) linearity, image contrast, spatial resolution, and absorbed dose for CBCT.

2.A | Vero4DRT
The characteristics of the Vero4DRT system were published previously 11 (Fig. 1). The gantry is extremely rigid owing to its O-ring shape. The Vero4DRT system has two kV x-ray imaging subsystems attached to the O-ring and two FPDs at 45°with respect to the MV beam axis. The CBCT images are acquired using kV x-ray tubes by rotating the gantry.
To acquire a set of SCBCT images, a kV X-ray source is rotated 215°clockwise (CW) or counterclockwise (CCW) [ Fig. 2(a,b)]. To acquire a set of DCBCT images, both kV x-ray sources are rotated 115°CW or CCW [ Fig. 2(c)]. In our study, SCBCT was performed using only one tube and DCBCT was performed using the two tubes simultaneously. We used only CW rotation because there would be no difference in image quality between CW and CCW rotations. 6 For image acquisition, we used the following x-ray parameters: tube voltage = 120 kV, tube current = 200 mA, and pulse width = 10 ms.

2.B.1 | Image uniformity
The uniformity module CTP 486, with a uniform disk, was used to assess image uniformity. were assessed (Fig. 3). Image uniformity was calculated using the following equation: where CT ROI;peripheral and CT ROI;center are the mean pixel value of the ROIs at the four peripheral positions and the center, respectively.

2.B.3 | Image contrast
We used the CTP 404 module, described above, to assess the image contrast, which was calculated using the following equation: where P ROI;insert is the mean pixel value in a circled ROI inside an insert and ROI, background is the mean pixel value of the background of the contrast module, respectively.

2.B.4 | Spatial resolution
The high-resolution module CTP 528 has a high-resolution pattern of 1 through 21 line pairs per centimeter (Lp/cm). High-contrast resolution was calculated using the method reported by Droege et al. 13 In that method, the practical modulation transfer function (pMTF) curve is calculated by measuring the standard deviation of the pixel values in each individual pattern in the cyclic bar pattern image. To assess the spatial resolution quantitatively, 50% and 10% values were calculated from the pMTF curve data.

2.B.5 | Absorbed dose
Dose measurements were performed using methodology adapted from that outlined in AAPM Task Group Report No. 111. 14 The CBCT absorbed dose was measured using a 0.6-cm 3  The average absorbed dose was calculated by analogy to the weighted cone-beam CT dose index (CBCTDI w ) 14 using the following equation: where D center is the central axis dose and D peripheral is the average peripheral dose of the scanning phantom. 15 The other holes of the phantom were filled in with PMMA rods to avoid affecting the measurements. The absorbed dose was measured five times for each position of the ionization chamber.      Table 2 shows the spatial resolution results for the 50% and 10%

3.D | Spatial resolution
MTF obtained with SCBCT (tube 1/tube 2) and DCBCT. There was no significant difference in spatial resolution between SCBCT and DCBCT.

| DISCUSSION
We compared the image quality and absorbed dose for the first commercial DCBCT, in the Vero4DRT image-guided radiotherapy system, with those for SCBCT. The CBCT of the Vero4DRT system cannot rotate a full 360°, which leads to lower image uniformity.
The image uniformity of DCBCT is worse than that of SCBCT. With DCBCT, the detector may detect the photons scattered by the object, resulting in degradation of the image quality. 16  Several authors proposed techniques to reduce the cross-scatter and improve the image quality for DCBCT [16][17][18]20]. Giles et al. 17 proposed that almost all cross-scatter effects can be removed by interleaved acquisition, which can be achieved at the same angular sampling rate by either doubling the data acquisition rate or halving the rotation speed. In another study, a bowtie filter was used to reduce the scatter-to-primary radiation ratio and improve image quality. 20 Because of the limitation of the hardware used in scatter-reduction methods, Zhu et al. 16  with respect to clinical patient data because a phantom is relatively uniform, unlike a patient. Figure 7 shows axial images of the pelvic region obtained with SCBCT and DCBCT. A CBCT correction algorithm for clinical use is necessary to improve the DCBCT image quality. 21 We will further investigate the performance of DCBCT in other imaging regions (e.g., thorax and abdomen).

| CONCLUSION S
In this study, we evaluated the image quality and absorbed dose for DCBCT by comparing them with the corresponding values for SCBCT, using the Vero4DRT. Compared with SCBCT, DCBCT had slightly reduced image uniformity. A strong linear correlation existed between the mean HU values in the ROIs obtained by DCBCT and the reference HU, although the linear regression slope was different from that of conventional CT. The image contrast with DCBCT, particularly for extreme-contrast material, was worse than that with SCBCT, even though the absorbed dose was higher.

ACKNOWLEDG MENTS
The authors thank Mr. Satoshi MURAKI of Mitsubishi Heavy Industries, LTD. for useful advice.

CONFLI CT OF INTEREST
No conflicts of interest.