CHAPTER 6 QUALITY ASSURANCE OF VARIAN ON-BOARD IMAGER

127 CHAPTER 6 QUALITY ASSURANCE OF VARIAN ON-BOARD IMAGER 6.1 INTRODUCTION Accurate and repeatable setup of patients is a requisite in radiotherapy. In the treatment of head-and-neck tumors, accurate setup is particularly important because of the proximity of the treatment volume to critical structures such as the spinal cord, brainstem, and parotid glands. When intensity-modulated radiotherapy (IMRT) is used, proximity of this kind between the target and normal organs often leads to highly inhomogeneous fluence profiles with steep dose gradients. Thus, repeatable and accurate patient setup is crucial to prevent the potential hazard of encompassing critical structures for example, the spinal cord in the high-dose region. Indeed, studies have shown that improper setup in the head-and-neck region can have significant effects on tumor coverage and normal-tissue sparing (Manning et al 2001, Samuelsson et al 2003, Astreinidou et al 2005). Improvement of head-and-neck setup precision and reproducibility therefore continues to be an active field of study. One approach to evaluating setup errors is to classify them into two categories: interfractional and intrafractional. The former reflect setup differences from day to day, and the latter, changes in patient position during a treatment session. Setup verification studies more often measure the former (Hess et al 1995, Pissani et al 2000, Asselen et al 2004, Kim et al 2004, Hong et al 2005, Guckenberger et al 2006, Linthout et al 2006, Zhang et al 2006), but some measure the latter (Kim et al 2000, Hong et al 2005, Guckenberger

128 et al 2006, Linthout et al 2006). Measurement of interfractional and intrafractional setup error through the course of treatment therefore requires repeated imaging of the patient on the treatment table. This imaging may result in excessive dose to the patient if megavoltage (MV) imaging is used. Kilovoltage (kv) imaging results in a lower patient dose and it can therefore be used with sufficient frequency to acquire more setup information per patient than traditional MV imaging allows. The advent of onboard kv imaging devices has also introduced new and more effective means for measuring and correcting setup error at the treatment machines. These devices not only allow measurement of setup error at the treatment console, but also remote correction of the setup based on the measured error. The advantages of kv over MV imaging for setup correction have been shown to include not only lower dose, but also smaller interobserver variability and significantly better setup error reduction in the headand neck region (Pissani et al 2000). Because these imagers are of the projection variety, little information about soft tissue is available, and registration occurs by alignment of bony anatomy. But the imagers can also be used to generate cone-beam computed tomography (CT) scans if soft tissue registration is required. The On-Board Imager system (On-Board Imager, Varian Medical Systems, Inc., Palo Alto, CA) is designed to correct for motion and setup errors of patients undergoing radiation therapy. The OBI system provides three imaging modes: two-dimensional (2-D) radiographic acquisition, fluoroscopic image acquisition, and three-dimensional (3-D) cone-beam computed tomography (CBCT) acquisition. The fluoroscopic images are used to verify the gating thresholds of the respiratory gating system (RPM, Varian Medical Systems, Inc.) to account for intrafraction (i.e., respiratory) motion. The radiographic images manage interfractional motion and setup

129 errors. The CBCT images provide soft tissue and bony structure information in 3D and also manage interfractional motion and setup errors (Jaffray and Siewerdson 2000, Siewerdson and Jaffray 2001, Guan and Lu 2005). Using the 2D2D Match and 3D3D Match analysis tools a user can register the acquired kv or CBCT images with their associated reference image (e.g., digitally reconstructed radiograph (DRR) or planning CT). Couch corrections are then downloaded to the linear accelerator and the couch is moved remotely. The use of this new technology necessitates a comprehensive quality assurance (QA) program to maintain and monitor system performance characteristics, which have been established at the time of commissioning. Currently, there have been no published recommendations and guidelines for a QA program to verify the functionality, accuracy, stability, and image quality of the radiographic and CBCT modes of this device. Several early adopters of the OBI technology independently developed their own QA programs using similar but not identical methods. This paper combines the best of these methods to generate a comprehensive QA program for the OBI and CBCT system. The QA program has three components: safety, geometry, and image quality for both radiographic and tomographic images. The most critical tests are those that evaluate the geometric accuracy of the OBI system, since the OBI system is intended for repositioning of the patient before/during treatment. All the geometric tests described in this publication assume that the location of the MV isocenter is known and can be identified independently (Kutcher et al 1994). Therefore, the tools used to identity the location of the MV isocenter (e.g., wall lasers, field light) are assumed to be calibrated. Image quality needs to be checked frequently enough to endure consistent image quality. Tests of image quality do not need to be as frequent as the geometry tests, but a regular program to detect changes in the imaging chain is still important.

130 6.2 REVIEW OF LITERATURE Yoo et al (2006) developed a quality assurance (QA) program for the On-Board Imager (OBI) system and to summarize the results of these QA tests over extended periods from multiple institutions. Both the radiographic and cone-beam computed tomography (CBCT) mode of operation have been evaluated. The QA programs from four institutions have been combined to generate a series of tests for evaluating the performance of the On-Board Imager. The combined QA program consists of three parts: (1) safety and functionality, (2) geometry, and (3) image quality. Safety and functionality tests evaluate the functionality of safety features and the clinical operation of the entire system during the tube warm-up. Geometry QA verifies the geometric accuracy and stability of the OBI/CBCT hardware/software. Image quality QA monitors spatial resolution and contrast sensitivity of the radiographic images. Image quality QA for CBCT includes tests for Hounsfield Unit (HU) linearity, HU uniformity, spatial linearity, and scan slice geometry, in addition. All safety and functionality tests passed on a daily basis. The average accuracy of the OBI isocenter was better than 1.5 mm with a range of variation of less than 1 mm over 8 months. The average accuracy of arm positions in the mechanical geometry QA was better than 1 mm, with a range of variation of less than 1 mm over 8 months. Measurements of other geometry QA tests showed stable results within tolerance throughout the test periods. Radiographic contrast sensitivity ranged between 2.2% and 3.2% and spatial resolution ranged between 1.25 and 1.6 lp/mm. Over four months the CBCT images showed stable spatial linearity, scan slice geometry, contrast resolution (_1%; < 7 mm disk) and spatial resolution (6 lp/cm). The HU linearity was within ±40 HU for all measurements. By combining test methods from multiple institutions, we have developed a comprehensive, yet practical, set of QA tests for the OBI system. Use of the tests over extended periods show that the OBI system has reliable mechanical accuracy and stable

131 image quality. Nevertheless, the tests have been useful in detecting performance deficits in the OBI system that needed recalibration. It is important that all tests are performed on a regular basis. Mechalokos et al (2007) used a kilovoltage imaging device to measure interfractional and intrafractional setup deviations in patients with head-and-neck or brain cancers receiving intensity-modulated radiotherapy (IMRT) treatment. Before and after IMRT treatment, approximately 3 times weekly, 7 patients were imaged using the Varian On-Board Imager (OBI: Varian Medical Systems, Palo Alto, CA), a kilovoltage imaging device permanently mounted on the gantry of a Varian 21EX LINAC (Varian Medical Systems). Because of commissioning of the remote couch correction of the OBI during the study, online setup corrections were performed on 2 patients. For the other 5 patients, weekly corrections were made based on a sliding average of the measured data. From these data, we determined the interfractional setup deviation (defined as the shift from the original setup position suggested by the daily image), the residual error associated with the weekly correction protocol, and the intrafractional setup deviation, defined as the difference between the post-treatment and pretreatment images. We also used our own image registration software to determine interfractional and intrafractional rotational deviations from the images based on the templatematching method. In addition, we evaluated the influence of inter-observer variation on our results, and whether the use of various registration techniques introduced differences. Finally, translational data were compared with rotational data to search for correlations. Translational setup errors from all data were 0.0 ± 0.2 cm, 0.1 ± 0.3 cm, and 0.2 ± 0.3 cm in the right left (RL), anterior posterior (AP), and superior inferior (SI) directions respectively. Residual error for the 5 patients with a weekly correction protocol was 0.1 ± 0.2 cm (RL),

132 0.0 ± 0.3 cm (AP), and 0.0 ± 0.2 cm (SI). Intrafractional translation errors were small, amounting to 0.0 ± 0.1 cm, 0.1 ± 0.2 cm, and 0.0 ± 0.1 cm in the RL, AP, and SI directions respectively. In the sagittal and coronal views respectively, interfractional rotational errors were 1.1 ± 1.7 degrees and 0.5 ± 0.9 degrees, and intrafractional rotational errors were 0.3 ± 0.6 degrees and 0.2 ± 0.5 degrees. No significant correlation was seen between translational and rotational data. The OBI image data were used to study setup error in the head-and-neck patients. Nonzero systematic errors were seen in the interfractional translational and rotational data, but not in the intrafractional data, indicating that the mask is better at maintaining head position than at reproducing it. The performance of a flat panel detector was investigated by Ejere (2006) for linearity, uniformity and bad-pixels. Especially at lower and higher exposures, the flat-panel detector was non-linear. The same exposures that cause non-linearity of the flat-panel detector also caused non-uniformity. The detector was linearized piecewise and exposures that cause non-uniformity were sought and calibrated. Pixel-by-pixel standard deviation investigation was used in correcting non-uniformity. They have compromised between the number of calibration points and the costly calculations involved. Bad-pixels were also examined and replaced by the average of the neighboring pixels. The effect of X-ray scatter on image quality was examined. X-ray scatter has the most detrimental effect on cone-beam CT image quality. Streaks and cupping are the most prominent artifacts caused by x-ray scatter. On top of that, scatter reduces contrast and increases noise. Scatter was estimated and corrected using three methods: Estimating scatter as uniform given by scatter-to-primary ratio (SPR) of 0.33, beam stop array technique, and simulation. Scatter was estimated by choosing a constant SPR of 0.33. This way of estimating and correcting scatter reduced the cupping artifact to

133 some extent but did not eliminate the artifact completely. To estimate scatter more accurately, scatter was sampled using beam stop array technique and interpolated to generate scatter-only image. Scatter-only image was then subtracted from the total image to give the scatter corrected primary image. The total image and the scatter corrected primary image were then compared in order to inspect the effect of scatter on image quality and evaluate the scatter correction methods. Effect of scatter on CT number accuracy, contrast and noise was investigated using an image quality phantom. The method of simulation considered the effect of X-ray source fluctuation, area occupied by the beam stoppers, transmission of the lead discs and transmission of the phantom itself. Inclusion of these factors in their beam stop array correction method produced considerable result. CT number inaccuracy or cupping artifact was corrected after the employment of the scatter correction methods. Contrast improved satisfactorily. However, all the scatter correction methods incurred noise. Knight (2006) used daily Computer Tomography (CT) imaging to comprehensively assess organ volumes, organ motion and their effect on dose. Daily CT images were obtained using a Siemens Primus Linear Accelerator equipped with an in-room Somatom CT unit in the accelerator suite, marketed as Primatom, to accurately position the patient prior to treatment delivery. The internal structures of interest were contoured on the planning workstation by the investigator. The daily volume and location of the organs were derived from the computer to assess and analyse internal organ motion. The planned dose distribution was then imported onto the treatment CT datasets and used to compare the planned dose to i) the actual isocentre, where the isocentre was actually placed for that fraction, ii) the uncorrected isocentre, by un-doing any on-line corrections performed by the treatment staff prior to treatment delivery, and iii) the future isocentre, by placing the isocentre relative to internal organ motion on a daily basis.

134 The results of this study did not confirm a statistically significant decrease in rectum volumes over time (hypothesis 1), however large fluctuations in bladder volume were confirmed (hypothesis 2). Internal organ motion for the rectum and bladder was demonstrated to be related to organ filling. Ideal planning volumes for these organs have been reported to minimise systematic and random uncertainty in the treatment volumes. An observed decrease in prostate volume over time, a systematic uncertainty in the location of the prostate at the time of the planning CT scan and a significant relationship between prostate centre of volume and rectum and bladder volumes has resulted in a recommendation that patients should be re-scanned during treatment to ensure appropriate clinical target volume coverage. A significant relationship between rectal and bladder volumes and the dose delivered to these organs was found (hypothesis 3). The dose delivered to the planning target volume was not related to the rectal or bladder volumes, although it was related to the motion of these organs. Despite these results only minimal effects on the dose delivered to any of the three isocentres occurred, indicating that the planned dose was accurately delivered using the methodology presented here (hypothesis 4). However the results do indicate that the patient preparation instructions need to be improved if margins are to be reduced in the future. Millos (2006) investigated to evaluate the performance of an On-Board Imager (OBI) mounted on a clinical Linear Accelerator. The measurements were divided into three parts, geometric accuracy, image registration and couch shift accuracy, and image quality. A cube phantom containing a radiation opaque marker was used to study the agreement with treatment isocenter for both kv-images and cone-beam CT(CBCT) images. The long term stability was investigated by acquiring frontal and lateral kv images twice a week over a 3-month period. Stability in vertical and

135 longitudinal robotic arm motion as well as the stability of the center of rotation was evaluated. Further, the agreement of kv-images and CBCT center with MV image center was examined. A marker seed phantom was used to evaluate and compare the three applications in image registration 2D/2D, 2D/3D and 3D/3D. Image registration using kv-kv images sets were compared with MV-MV and MV-kV image sets. Further, the accuracy in 2D/2D matches with images acquired at non-orthogonal gantry angles was evaluated. The image quality in CBCT images was evaluated using a Catphan phantom. Hounsfield unit (HU) uniformity and linearity was compared with planning CT. HU accuracy is crucial for dose verification using CBCT data. The geometric measurements showed good long term stability and accurate position reproducibility after robotic arm motions. A systematic error of about 1 mm in lateral direction of the kv-image center was detected. A small difference between kv and CBCT between kv and MV images center was > 2 mm at some gantry angles. Image registration with different match applications worked sufficiently. 2D/3D match was seen to correct more accurately than 2D/2D match for large translational and rotational shifts. CBCT images acquired with full-fan mode showed good HU uniformity but half-fan images were less uniform. In the soft tissue region the HU agreement with planning CT was reasonable while a larger disagreement was observed at higher densities. This work shows that the OBI is robust and stable in its performance. With regular QC and calibrations the geometric precision of the OBI can be maintained within I mm of treatment isocenter. Different aspects of the imaging system (called XVI) of Elekta Synergy machine has been evaluated by Renstrom (2005), such as image quality, flex of the gantry rotation, accuracy for patient setup correction, x-ray tube output and absorbed imaging doses. The stability in 2D and 3D image

136 quality as well as the mechanical stability during rotation (also called flex) was measured on a weekly basis during a three month period. The stability of the system was studied by imaging a dense sphere positioned at isocenter. The 3D imaging was evaluated with a Catphan 500 phantom and the 2D imaging with a Leeds phantom. Accuracy in patient setup corrections was evaluated by scanning a pelvis phantom in different locations. Readout from the software was compared to actual translational movements, which were defined by the room laser readouts on rulers mounted on the table-top in the x- y and z-direction. A gold marker matching study was performed using an Alderson/Rando phantom. Finally, both CT dose index (CTDI) and dose measurements with TLDs were performed. The TLDs were placed between the slices of the Alderson/Rando phantom using thin boards of polystyrene. The XVI showed minor long term variations in flex and stable image quality. The accuracy of detecting patient setup-errors based on bony structures was flawless, although matching with respect to gold markers failed. The received effective dose from a pelvis scan was about 11 msv and for a head-and-neck scan about 0.15 msv. Accurate and reproducible setup of patients is a important part in radiotherapy. In recent development the IGRT with On Board Imaging System helps to achieve this goal. A series of QA tests were performed with equipments and tools as recommended by Varian and as per their QA protocols OBI customer acceptance procedure and OBI CBCT customer acceptance procedure. The results of the OBI QA procedures were within the specified limits as per Varian protocol. This test should be carried at a specific regular frequency to ensure the quality use of OBI for IGRT.

137 6.3 RESULTS Accurate and reproducible setup of patients is a important part in radiotherapy. In recent development the IGRT with On Board Imaging System helps to achieve this goal. A series of QA tests were performed with equipments and tools as recommended by Varian and as per their QA protocols OBI customer acceptance procedure and OBI CBCT customer acceptance procedure. The results of mechanical test are given in Tables 6.1 to 6.9. Table 6.1 Mechanical Tests and Results Sr. No. Test Specification Observation Comments Visual and Sound Collision test Motion Enable Test Mechanical isocenter Door interlock Warning light Audio sound warning KVS Collimator Cover KVD Imager Cover KVD/KVS Arm Paddles Pendant Manual Motion Console Automatic Motion Collimator Rotation Gantry Rotation radius radius

138 Table 6.1 (Continued) Sr. No. Test Specification Observation Comments Source Vertical Position vs PRO Tube Warm-up Technique PRO Vertical Postion Actual vertical Position Spec 100.0 cm (P1) 99.95 ±2mm 80.0cm (P2) 79.95 Vertical Travel Range Pulsed Fluoro Mode Full Travel Range 21.0 cm kv ma ms Beam-on Time 75 80mA 32ms 20-30 seconds Tool Measurements on a 10 x 10 cm image Imager Panel Alignment Optical Isocenter Test Tool Width Height spec Measured Distance 99.9 99.9 100mm±2mm Measured Area 99.86 99.99 100mm±2mm Measured Angle 45º 45º±1.15º Pixel position Actul Specification KVS/KVD Isocenter 1024x764 Actual 0.3 mm 1024x768±10(±2.0mm) Specification

139 Table 6.2 Imager Positioning Vs. Pro and Travel Ranges PRO VERTICAL POSITION 0.0 cm (P5) -30.0 cm(p4) -50.0 cm (P2) -75.0 cm (P1) -50.0 cm (P2) Vertical Travel Range Vertical Position Longitudinal Travel Range Lateral Travel Range Actual (add No spec No Spec offset) 1mm Spec±2mm Actual (add Actual Range offset) No Spec 30.0-20.9 24.6 Spec±2mm Spec -20.5cm to =24cm) Actual (add Actual Range Actual Ranger offset) 49.95-23 +24.6-19.8 +16.1 Spec±2mm Spec -19cm to Spec -18cm to =24cm) =16cm) Actual (add Actual Range Actual Ranger offset) 75.0-19.8 +24.6-19.8 +16.1 Spec±2mm Spec -19cm to Spec -18cm to =24cm) =16cm) Longitudinal Position at Longitudinal Position at +10.0 cm -10.0 cm PRO Actual PRO Actual + 10.0 + 10.5-10.0-10.5 Spec±2mm Spec±2mm Lateral position at +10.0 Lateral position at -10.0 cm cm PRO Actual PRO Actual + 10.0 + 10.5-10.0-10.5 Spec±2mm Full Travel Range 85.5 Spec±2mm Spec: 80.0 cm below isocenter)

140 Table 6.3 Digital Fluoroscopy kvp, ma and ms measurements Indicated kvp Indicated ma Pulsed fluoro Technique Indicated ms Actual kvp Actual ma Acutal ms 60 25 30 60.2 23.3 29.6 60 50 20 60.4 48.2 29.6 80 50 6 79.7 48.9 5.6 80 50 15 79.7 48.2 14.7 80 50 20 79.7 48.2 19.8 90 25 30 89.9 23.7 29.6 90 80 15 89.9 76.1 14.6 120 25 10 119.6 24.5 9.5 120 80 10 119.3 48.2 9.5 Table 6.4 Digital Fluoroscopy kvp, ma and ms measurements Indicated kvp Sinlgle Standard-Single pulse half resolution technique Indicated ma Indicated ms Actual kvp Actual ma Acutal ms 60 25 30 59.8 24.5 29.4 60 50 40 59.8 48.2 39.2 80 50 40 79.3 48.2 39.2 80 50 50 79.3 48.5 49.4 80 50 100 79.3 48.2 99.5 80 50 150 79.3 48.2 149.5 90 25 200 89.5 24.1 199.7 90 80 100 89.5 76.9 99.2 120 25 50 119.3 24.5 50.0 120 80 25 119.2 47.8 24.4

141 Table 6.5 Half value layer (HVL) using digital Fluoroscopy Pulsed Mode Unfiltered Pc Filtered Pc mm of Aluminum 70 kvp 301.4 150.52 2.7 100 kvp 602.5 300.99 2.03 Table 6.6 Shows Density resolution (Hounsfield/CTNo.) Standard dose 150cm Bow Tie Mean CT- No. Tolerance Measured CT-No. Head Scan Air 1000 ± 40-999 Acrylic 120 ± 40 109 LDPE 100 ± 40-103 Body Scan Air 1000 ± 40-995 Acrylic 120 ± 40 104 LDPE 100 ± 40-126 Table 6.7 Shows the Spatial Linearity measurement Standard dose 150cm Bow Tie Distance Accuracy Catphan Performance Measurement Head Scan 50 mm ±1 % 49.8 Body Scan 50 mm ±1 % 49.7

142 Table 6.8 Shows the High resolution Standard dose 150cm Bow Tie Line Pair / cm Specification Actual Line Pair / cm row Head Scan Row 7 8 Body Scan Row 6 7 Table 6.9 Shows the Low Contrast Resolution Standard dose 150cm Bow Tie Specification Supra Slice % Specificatio n Target size Actual Supra Slice % Actual Target size Head Scan 1 % 15 mm 1% 6 mm Body Scan 1 % 15 mm 1% 6 mm 6.4 DISCUSSION Accurate and reproducible setup of patients is a important part in radiotherapy. In recent development the IGRT with On Board Imaging System helps to achieve this goal. A series of QA tests were performed with equipments and tools as recommended by Varian and as per their QA protocols OBI customer acceptance procedure and OBI CBCT customer acceptance procedure. The results of the OBI QA procedures were within the specified limits as per Varian protocol. This test should be carried at a specific regular frequency to ensure the quality use of OBI for IGRT.